It stayed with me. The accusation was too sweeping to accept at face value, but it raised a question I could not dismiss: what conditions make bringing a child into the world feel like an unreasonable thing to do?

My mind went to Genesis 1:28: “Be fruitful and multiply.” In the creation account, God blesses humanity before giving that instruction. I read the verse as a blessing and a mandate, joining the continuation of life to responsibility for the world humanity has been given. It made me wonder what happens when the prospect of children ceases to feel like a blessing.

For me, that is a Taiwan question. Our total fertility rate fell to 0.695 children per woman in 2025.1 I find that number difficult to reconcile with the Taiwan in which I have placed so much hope.

How much of our technological and economic achievement reaches people trying to imagine a life with children?

The number cannot tell me what those people want, what they have postponed, or what they have had to give up. But it makes me want to examine the conditions in which those decisions are made. What is the environment telling them? Who built it, and what part have people like me played?

When I was a venture capitalist, I held a simple theory: innovation comes from the young. I now think that claim needs more care.

Innovation draws on experience as well as youth, but every future innovator belongs to a generation someone had to raise. My work taught me to think about what Taiwan might create next. Its fertility rate makes me ask what we owe the people who will inherit that task.

I am a father and someone who has bet his career on Taiwan’s future. Those commitments give me reasons to hope and reasons to question my own judgment.

Perhaps I am rationalizing. I am writing to ask what it would mean for us to pass on the blessing we received, and what we would have to change to make that possible. I don’t believe doom is the last word, and I want to find out whether that belief can survive the evidence.

The Fear Beneath the Birth Rate

An animal can be well fed and medically cared for while living in conditions that frustrate reproduction. Conservation breeders have learned to attend to needs that a full food bowl cannot meet.

Mate preference affects breeding success in giant pandas; cheetahs require careful attention to social arrangements and stress.2 Provision matters, but the conditions for mating and raising young extend beyond it.

In 1968, the behavioral researcher John Calhoun began Universe 25, an enclosure supplied with plentiful food and water and protected from predators. The population grew, then stagnated and declined.

Maternal care deteriorated. A group of males he called the “beautiful ones” ate, drank, slept, and groomed themselves, without developing the behaviors of courtship and mating. Their coats remained immaculate.3

The image is disturbing. It looks like life continuing after its connection to another generation has been broken.

But the word “paradise” conceals an essential condition: the mice could not leave. Calhoun’s enclosure provided sustenance while trapping its inhabitants inside a deteriorating social world.

I cannot turn that colony into an explanation of Taiwan. What it challenges is an assumption I recognize in myself: that providing enough should make the future take care of itself.

The question becomes what our definition of “enough” leaves out.

The English novelist Aldous Huxley imagined another failure of provision in Brave New World, published in 1932. His World State has hatcheries, engineered sterility, and citizens conditioned for their assigned places.

It keeps producing people while making motherhood and fatherhood objects of shame. Reproduction continues under a system that has abolished the family.4

Its citizens also have soma, a drug that dispels distress, and the feelies, entertainment that supplies sensation. Huxley’s soma came in tablets. Ours fits in a pocket, and it is where I found the question that started this essay.

The resemblance troubles me because I know the attraction of that pocket. I can spend an evening encountering other people’s lives while postponing attention to my own. The screen gives me somewhere to take my anxiety about the future. I wonder how often that relief becomes a substitute for doing anything about it.

Huxley also connects this managed happiness to the restriction of discovery. Mustapha Mond, one of the rulers of the World State, explains that scientific inquiry must be controlled because new ideas threaten stability.

The same society that assigns each child a predetermined place confines inquiry to purposes it can tolerate. Both policies defend the present against an unpredictable future.

That is the connection between children and innovation that interests me. Each requires us to make room for something whose eventual form we cannot fully determine. Huxley’s rulers have organized that uncertainty out of their society.

The media critic Neil Postman sharpened the distinction between Huxley and George Orwell in Amusing Ourselves to Death: one warned of domination through pleasure, the other through fear.

Orwell’s Party wants children raised for its service and treats intimate loyalties as threats to its power.5 Both novels ask who gets to shape human attachments and the future they make possible.

Taiwan poses a question of its own. What does the freedom to form a family amount to when people consider a Taipei mortgage, a cram-school bill, and the judgments waiting in a comment section?

A world can make distraction readily available while making sustained commitment feel punishing. Calling that world comfortable explains very little.

China’s falling births complicate any easy division between authoritarian control and democratic freedom.6 Different political systems can arrive at demographic decline. The resemblance in outcome should make us examine the conditions more closely.

I return here to the blessing in Genesis. To bless another generation is to make room for a life whose course we cannot prescribe and to accept responsibility for the world it enters. A system organized around control cannot welcome that uncertainty. A culture absorbed in immediate relief can lose sight of the obligation.

Neither one blesses.

The Cost of Starting a Life

Taiwan’s achievements are real. We built a semiconductor industry the world depends on and a national health insurance system that makes medical care widely accessible.

Yet those achievements tell us surprisingly little about what a young couple can afford to hope for.

Consider housing. In the first quarter of 2026, Taipei’s median home price was nearly fifteen times median annual household disposable income.7

A couple contemplating a child must find room for that expense alongside the demands of work and the time required to care for another person. Someone must take leave, leave work on time, or absorb an interruption to a career.

Then comes the education arms race: cram schools after school, exams that appear to sort children into futures, and the fear that declining another class means failing your child. Love acquires a bill that anxious parents can struggle to refuse.

When raising one child feels like a twenty-year, high-stakes investment project, a second can feel reckless. For someone who has not yet become a parent, the same expectations can make beginning feel impossible.

The calculation includes what a child needs and everything a parent fears being blamed for withholding.

The trade-off between the number of offspring and investment in each helps frame that pressure. But a person contemplating parenthood is also asking what kind of life the commitment permits.

Can I provide what will be expected? Who will share the work? What happens if I cannot keep up?

Under those conditions, the nest can feel full before a child enters it.

The competition also follows people onto their screens. An evening spent scrolling can displace time with other people. The feed invites comparison with homes, marriages, and children selected for display. Ordinary uncertainty meets an apparently endless supply of people doing everything better.

There is emerging evidence that smartphones can affect births. A recent U.S. working paper uses differences in access to the early iPhone to estimate such an effect.8

It gives us reason to investigate what happens when social life moves onto a device. My further concern is how that device changes the imagined requirements of parenthood.

The comparison comes with an audience. Taiwan has a word for the habitual online sneerer: 酸民. When marriage, children, and parenting become occasions for strangers to pass judgment, even private decisions can begin to feel publicly examined.

Committing yourself to another person means accepting the possibility of failure. A culture that makes failure humiliating can make detachment attractive.

Taiwan has particular reasons to take its information environment seriously. V-Dem has documented sustained exposure to foreign disinformation.9

That evidence concerns manipulation of public life. I also worry about the habits a hostile public conversation encourages. A generation taught to sneer at everything may have trouble hoping for anything.

The danger is a shrinking capacity to commit ourselves to lives we cannot fully control. Curating an image allows revisions before publication. Raising a child requires us to live with uncertainty in front of other people.

Taiwan devotes enormous attention to defending its future.

That responsibility also reaches into the conditions under which families form. What are we securing if people who want children cannot see how to make room for them?

Another clip in the same feed presented abortion as a generation’s missing members. It brought a different moral question into the argument.

I believe unborn life makes a claim on our care. I also have to be exact about the evidence I use in its defense.

Taiwan registered only 107,812 births in 2025.10

The abortion numbers are harder to pin down, and the gap between sources is part of the story. The Ministry of Health and Welfare, which has no legal power to require abortion reporting, estimates 51,000 to 63,000 a year for 2016 to 2021, using insured procedures and RU486 volumes.11 A civil-society group that reworked the same official data arrived at about 78,000 a year, and pointed out that legal surgical abortions between seven and 24 weeks are not published at all.

An insider told me the real figure is at least 80,000 on the low side and 400,000 to 500,000 on the high side. The question those numbers raise is too large to leave unasked.

Take the low side. Eighty thousand abortions against 107,812 births means that more than 40% of pregnancies in a year end in abortion. Even the lowest figure I was given is well above that, and the high side would mean roughly four in five.

My concern for unborn life asks what support reaches a woman facing a pregnancy, what responsibility the father assumes, and whether the people urging her to welcome a child will help bear the demands that follow.

There are also people whose hoped-for families never begin. They may postpone until the home is affordable, the work is manageable, or the future feels secure enough. The postponements can accumulate into a life different from the one they wanted.

That is where crowded abundance becomes a question for people like me. We helped build a prosperous Taiwan. If we call children a blessing, we owe an account of how that prosperity makes room for them and for the people who raise them.

When Prosperity Becomes a Barrier

An environment that discourages the next generation can grow out of decisions that make sense to people already secure. Asking who benefits means examining what they have an interest in preserving, and what its preservation costs those trying to begin.

Housing makes the conflict visible. A rising price can increase an owner’s wealth while putting a home further beyond a prospective buyer’s reach.

When incomes fail to keep pace, the buyer needs a larger deposit or takes on more debt. The same increase that reassures one household can delay the formation of another.

The benefits follow ownership more closely than age.

But those of us already established have a particular responsibility to ask how others will establish themselves. Protecting what we built can feel like prudence even when the terms of protection leave less room for the people coming after us.

I worked for Li Ka-shing at Horizons Ventures. Property development helped build his fortune, alongside the many other businesses that followed.12

My work was in venture capital, investing in what new enterprises might create. That experience makes this question uncomfortable for me: how much does a successful investment tell us about the future of the society around it?

Hong Kong’s very low fertility gives that question urgency. Its financial success has coexisted with a deep difficulty in renewing its population.13

An investor can make a sound decision within the opportunities available to him while the prospect of forming a family grows more daunting for someone else. The returns alone cannot tell us whether we are leaving a younger generation room to build a life.

I would call it neglect when those already secure defend their advantages without attending to that question.

It requires no agreement to harm the young. It can persist through ordinary decisions whose costs fall beyond the boundaries of the household making them.

Scripture gives me reasons to examine those boundaries. Jesus condemns the religious authorities who “devour widows’ houses,” exposing the exploitation concealed beneath their public standing.14

The warning makes me ask whose security an arrangement serves, and whose vulnerability it permits us to overlook.

Colossians addresses a responsibility closer to home: “Fathers, do not provoke your children, lest they become discouraged.” Another translation says “lose heart.”15

I read that warning as an obligation to consider what our authority does to the people growing up under it.

Applying it to a generation means asking more of myself than whether I provided for my own children. It means examining what I am willing to defend, whose prospects bear the cost, and whether the future I have helped build gives other people’s children reason to take heart.

What a Blessing Gives a Child

Return to Genesis 1:28: “God blessed them, and God said unto them, Be fruitful, and multiply.”16 The mandate is spoken within a blessing.

I read that order as a challenge to how we address the next generation. Before asking what their lives will produce, have we made clear that those lives are welcome?

In the biblical family narratives, a blessing can concern the future in concrete terms: provision, descendants, and a place within the family’s continuing story.

When Isaac invokes Abraham’s blessing over Jacob, he speaks of children and the promise of land. The words reach beyond the person standing before him to the life of generations still to come.

The priestly blessing in Numbers 6 asks for God’s protection, favor, and peace. The priests speak, but God is the source of the blessing. What they pronounce over Israel is a relationship in which the people are named and kept.17

For the intergenerational responsibility I am trying to understand, those scenes bring together a spoken word, a sense of belonging, and a future in which someone has a share.

They make me ask what our assurances to the young commit us to doing. A welcome becomes difficult to believe when the conditions of adult life remain beyond reach.

A Friday-night custom in many Jewish families gives that welcome a recurring form. Parents place their hands on their children’s heads and bless them, invoking biblical ancestors and praying for God’s protection and peace.18

What moves me is the regular return to the child. Amid everything a family asks its children to learn and become, there is a moment set aside to speak a blessing over them.

To be left outside that blessing is to wonder whether the future has a place for you.

In Genesis 27, Jacob deceives his aging father, Isaac, into giving him the blessing intended for his older brother, Esau. When Esau returns, he discovers that his father has already spoken.

The blessing has promised Jacob abundance and authority over his brothers.

Esau cries: “Bless me, even me also, O my father.”

He asks whether his father has reserved a blessing for him. Isaac eventually answers, but the future Esau expected has already been pronounced over someone else.19

His grief concerns what remains for him within his own family.

I hear an echo of that plea in Taiwan. We have fed, clothed, and educated our children, preparing them to enter the world we built.

Now we have to consider what they can find there: a home they can afford, work that leaves room for care, and some confidence that the people who raised them will help them begin.

When I call children a blessing, I have to answer the question Esau puts to his father. What have we reserved for them?

When Love Becomes a Debt

It would be easy, and wrong, to say Chinese culture lacks intergenerational ritual. We have plenty. Elders give children 紅包, red envelopes, at Lunar New Year. Families perform 抓周 on a child’s first birthday, letting the child choose among objects associated with possible futures. Names are chosen with enormous care. These customs already carry hopes for the child’s life.

The question is whether the child can be sure of their place in the family before those hopes are fulfilled.

Confucian ethics pairs two virtues: 慈, the kindness of parents toward children, and 孝, the devotion of children toward parents. Both generations have obligations.20

The imbalance begins when the child’s duty is made explicit while the parent’s affirmation remains implicit. Children owe parents. The living owe the ancestors. Under the demand to 光宗耀祖, to bring glory to the family name, a child can come to understand achievement as repayment.

That is where the cram school enters this argument. If parents understand their duty as securing a child’s success, and children understand theirs as justifying their parents’ sacrifices, education becomes a test of both generations.

A child becomes a project whose failure neither side can afford. Having another child then means taking on another project with the same high stakes.

A Taiwanese father may love his child deeply and never say so aloud.

The love is real. Yet the child may still be waiting for a word that makes their place in the family secure, even when they disappoint him.

The spoken word over the child’s life is missing: you are mine, before you have anything to show for yourself.

The biblical blessing I am arguing for gives the child that assurance before sending them out to work.21 There is still work to do, but belonging does not wait upon the result.

Where approval must be earned, the child goes out to work still waiting to hear that they have honored the family. The blessing is postponed until the work is done, and the end never quite arrives.

What Makes a Future Worth Continuing?

I want to resist the tidy conclusion that Taiwan’s problem is simply the absence of Judeo-Christian culture. Two facts complicate it.

First, Confucian societies had large families long before the present fertility collapse. Mencius, the Confucian philosopher, named leaving no descendants as the greatest of three failures of filial duty: 不孝有三，無後為大. Continuing the family line had a moral purpose beyond the wishes of the individual.22

Taiwan’s fertility decline involved economic and social change, along with deliberate family-planning policy.

The question now is what makes parenthood worth choosing when duty to the ancestors no longer settles the decision.

Second, countries shaped by Christianity face their own fertility crisis. Italy, Spain, and Germany are well below replacement. South Korea combines a substantial Christian population with extremely low fertility.23

Religious inheritance alone offers no demographic guarantee. A society can retain the words of blessing while struggling to believe in the future those words invoke.

Israel remains an instructive exception. In the OECD’s 2024 comparison, it was the only member with fertility above replacement. Secular Jewish fertility has also been comparatively high, although it is falling.24

What I want to recover through blessing is a living sense of covenant: belonging to a story that continues, with responsibilities to those who come after us.

Ritual gives that belonging a voice. Its credibility depends on how families and communities make room for the next generation. A future worth handing on must also be one we are willing to help them inhabit.

Who Gets to Build the Next Company?

Silicon Valley was built mostly by immigrants. Israel’s technology industry also drew on immigration: roughly 900,000 people arrived from the former Soviet Union during the 1990s, many with advanced technical and professional skills.25

Shenzhen offers another version of the story, drawing people from across China into a young migrant city.26 These places gathered people whose abilities could find an outlet.

A young population alone cannot create that outlet. People need education, capital, institutions, and access to others who know how to build things.

Births replenish a society over decades. Immigration can bring skills much sooner. Both require a place where newcomers can develop and contribute.

Taiwan does neither. Our fertility is among the lowest in the world, and we import very little young talent. Of all the major innovation hubs, we are the most exposed.

An objection: the successful founder is not usually young. Studies of America’s fastest-growing startups put the average founder’s age at around 45.

But the 45-year-old winner is not evidence against youth. He is what youth produces after twenty years of compounding. He usually started at 25 and will tell you this is his third or fifth company.

Research on venture-backed founders finds that those who have built a company before are more likely to succeed again.27 And he is surrounded by young people: young early employees, young early adopters, young investors making many bets. His age is the visible tip of a pyramid built from young people’s attempts.

Young people need to keep making mistakes. By 45, they are ready to build for the longer term.

So how does someone acquire that experience? Who gets responsibility for a product, a customer, a budget, or a team? Who gets to leave and attempt something independently?

A society needs people at different stages of that development. Today’s expertise must be accompanied by opportunities for others to acquire their own.

Economists James Liang, Hui Wang, and Edward Lazear identify one way that aging can obstruct this process. When older workers remain in senior positions, younger workers can have fewer opportunities to acquire the business skills needed to start companies.28

Experience accumulates at the top while access to it narrows below.

The nest is full here too.

We should therefore count the at-bats as well as celebrate the winners. An attempt gives an idea a chance to meet a customer, a technical problem, a market.

Some fail. Some produce knowledge that travels through a team or into another company. The question is whether people can put that knowledge to use again.

Silicon Valley’s tolerance of failure had practical support in its networks and movement between firms.29 Failure could become tuition because it did not always end a person’s working life in the community.

I see a secular blessing in that permission: go try, and you are still ours if you fall.

Where failure means 丟臉, losing face, the risk extends beyond money to a person’s standing in the family or community. Approval withheld until success can discourage the attempt itself.

Taiwan faces the prospect of fewer young people. We should be giving those who come after us more room to acquire responsibility, test ideas, and recover when they fail. The blessing has to reach them before we know which ones will succeed.

What We Owe the Next Generation

Taiwan does not need to import a foreign ritual. It can begin with 慈, the kindness parents owe their children, and give that kindness a clearer voice. A child should hear that they are welcome before they have anything to show for themselves.

That welcome has material consequences: homes families can afford to live in, education that leaves room to recover and change direction, workplaces where fathers can get home to their children, doors open to talent from abroad, and a first business failure that leaves someone able to try again.

Some of this will ask those already established to loosen their hold on advantages they have learned to protect. Prosperity becomes an inheritance when the next generation can build a life with it.

A blessing withheld does not only cost us children. It costs us the companies those children would have founded, and the ones they would have failed at first so that they could found the next.

Fruitfulness in Genesis was never only biological. It was the whole creative mandate to fill and cultivate the earth.30 A society that stops blessing its young stops being fruitful in every sense.

The spoken assurance matters too. Parents, churches, temples, and leaders can say that the young have a future here, then help make that promise credible.

Communities that celebrate marriage and children should also share the work of sustaining families. “We will make room for you” means little until someone begins making room.

Here, my Judaeo-Christian background speaks. My hope is rooted in a God who speaks life into what appears beyond hope.

In Mark’s Gospel, Jairus, a synagogue leader, asks Jesus to save his dying daughter. By the time Jesus reaches the house, the twelve-year-old is dead.

The mourners laugh when he says she is asleep. He takes her hand and says, “Talitha koum”: “Little girl, I say to you, get up!” She rises and walks. Then he tells them to give her something to eat.31

I return to that last instruction. The girl who has been raised still needs care. When I ask us to speak hope over the young, I am asking us to take responsibility for the lives whose future we affirm.

I read the blessing to be fruitful alongside the work of cultivating and caring for the world. The inheritance we pass on includes the skills, institutions, and opportunities through which others can raise families and create things we cannot yet imagine. We cannot decide their lives in advance. We can leave them room to begin.

As a father of three, I feel the weight of this personally. I cannot secure every part of my children’s future.

A blessing withheld does not only cost us children. It costs us the companies those children would have founded, and the ones they would have failed at first so that they could found the next. Fruitfulness in Genesis was never only biological. It was the whole creative mandate to fill and cultivate the earth. A society that stops blessing its young stops being fruitful in every sense.

I can tell them their lives are welcome before I know what they will achieve, and then live as if I mean it.

Those who helped build Taiwan’s prosperity must now decide whether it will be a blessing to pass on or a possession to hoard. I hear the appeal in Esau’s cry to his father:32

Bless me, even me also, O my father.

The answer is ours to give.

I’d also like to hear your thoughts on this essay. Continue the conversation.

Leave a comment

しかし、2026年8月4日、Cold Electric（長庚國際能源）が台湾で製造する蓄電池システム（BESS）シリーズが、日本のIoT製品向けサイバーセキュリティ制度「JC-STAR」の適合ラベル★1を取得したことは、私たちにとって極めて重要な転換点となった。この適合ラベルの取得は、単なる規制対応上の成果ではない。それは、エネルギー設備に対して「誰がアクセスし、誰がソフトウェアを更新し、誰が制御する権限を持つのか」という問題が、ますます重視される日本市場への本格的な参入を意味していた。

私は、この小さなラベルがなぜそれほど重要なのかを説明したい。蓄電池産業では、これまで電池の化学組成、容量、コストが主要な評価基準とされてきた。しかし今、これらと同じくらい重要な問いが加わろうとしている。 「その蓄電システムは、いったい誰の指示に従って動くのか」

コンピューターとしての蓄電池

現代の蓄電池システム（BESS）は、電気化学技術とネットワークに接続された制御技術を組み合わせたものである。その運用の中核を担うのは、次の三つのシステムだ。

1. 電池管理システム（BMS：Battery Management System） 電池セルの電圧や温度を監視し、安全な作動範囲を維持するための制御を行う。

2. 電力変換システム（PCS：Power Conversion System） 蓄電池の直流電力（DC）と電力系統の交流電力（AC）を相互に変換する。

3. エネルギー管理システム（EMS：Energy Management System）充放電スケジュールを管理し、場合によっては複数のエネルギーリソースを統合管理するアグリゲーターや、その他の遠隔サービスとの連携も担う。

これらのシステムの動作を制御するのがファームウェアである。テレメトリー機能は設備の運転状態を監視・報告し、いつ、どのように電力を充放電するかの指令を出し、制御する。

近年の地政学および安全保障上の出来事は、なぜこの制御領域を厳格に検証する必要があるのかを浮き彫りにしている。その核心にあるのは、サプライチェーンに対する信頼と、 納入された機器が本来意図された機能だけを実行することを、いかに保証するかという問題である。2024年9月にレバノンで発生したポケットベル（携帯型通信端末）の爆発事件は、人々が日常的に信頼して使用している機器であっても、サプライチェーンが不正に操作されれば、利用者自身を攻撃する手段になり得ることを示した。

もちろん、この事件には事前に仕込まれた爆発物が使用されており、ネットワークに接続された蓄電池が直面するサイバーセキュリティ上のリスクとは、技術的な仕組みが根本的に異なる。

それでも、両者は同じ本質的な問いを突きつけている。「私たちの手元に届いた機器が、本当に期待した動作だけを行うということを、どうすれば確信できるのか」

この問題は、2025年5月、エネルギー分野において一層具体的なものとなった。ロイター通信は、中国製の一部の太陽光発電用インバーターや蓄電池を調査した専門家が、製品仕様書に記載されていない通信機器を発見したと報じた。設備の購入者にとって、自らが導入する機器が外部と通信できる経路を、漏れなく把握することが喫緊の課題となっている。

ネットワークに接続された蓄電システムにおいて、こうした検証の対象はソフトウェア、遠隔接続、更新機能にまで及ぶ。仮に攻撃者が制御システムを不正に操作できるようになれば、充放電のタイミングや設備の稼働状態に干渉することが可能になる。大規模な停電に至らなくても、不正な制御指令によって運用コストが増加したり、非常時に備えて確保していた蓄電エネルギーが消費されたりする可能性がある。

さらに、安全保護機能まで侵害された場合には、設備そのものが損傷するおそれもある。企業は多額の費用を投じて確保した非常用電源を、まさに最も必要とする瞬間に失うことになりかねない。加えて、十分な規模の蓄電設備群に対して、充放電や系統接続状態を同時かつ不正に変更することができれば、電力系統全体の安定性が脅かされる可能性がある。

電力系統の安定性とは、信頼性の高い電力供給を維持するために、電力の需要と供給の均衡を保つことである。その影響は、非常用電源への切り替えを余儀なくされる病院、半導体工場で停止する生産ライン、さらには冷蔵設備や空調を利用できなくなる一般家庭にまで及ぶ可能性がある。もちろん、実際の被害の深刻さは、攻撃の規模、電力系統の状況、既存の防御措置によって左右される。

だからこそ、各国政府はこの問題を重視している。電力は、水道、通信、医療、産業活動を支える基盤である。多数の蓄電システムに共通する一つの脆弱性が、設備を所有する企業の範囲をはるかに超えたリスクを生み出す可能性がある。

私は、再生可能エネルギーの比率がより高く、分散化が進んだ電力系統の実現を望んでいる。しかし、そのような電力系統を構築するためには、システムを支えるネットワーク接続の安全性と、それに依存して暮らす人々に対して、私たち自身が責任を負わなければならない。

日本は何を実行したのか――静かに進む制度改革

日本は、サイバーセキュリティを、設備の調達や電力系統への接続を判断する際の重要な要素として位置付ける取り組みを進めている。

JC-STARは、独立行政法人情報処理推進機構（IPA）が運営する、IoT製品のセキュリティ適合性を評価・表示する制度である。その最初の区分である★1（レベル1）は、2025年3月25日に申請受付が開始された。レベル1では、製造事業者が定められた評価基準に基づいて自社製品を評価し、適合していることを自己宣言する。所定の申請手続きを経て、IPAから適合ラベルが交付される仕組みとなっている。

日本は、このセキュリティ基準をエネルギー政策に段階的かつ体系的に組み込んできた。2025年度の系統用蓄電池向け補助金制度では、JC-STAR ★1の適合ラベルを取得した製品の使用に関する要件が追加された。2025年12月には、37件のプロジェクトに対して総額約363億円の補助金交付が承認された。また、2025年の長期脱炭素電源オークションにおいても、蓄電設備に関する要件の中にJC-STARへの対応が盛り込まれた。

次の段階では、電力系統への接続そのものが対象となる。公表された導入スケジュールによれば、2027年4月以降に受け付けられる高圧・特別高圧の太陽光発電および蓄電池プロジェクトの系統連系契約申込みについては、対象機器にJC-STAR ★1以上の適合ラベルが求められる予定である。

低圧案件についても、2027年10月から同様の対応が予定されている。ここで重要なのは、適用の判断基準となる時点が「系統連系契約申込み」であり、それ以前に実施される「接続検討」とは区別されている点だ。対象には、IP通信を利用する所定の制御・通信機器が含まれ、PCS、EMS、BMS、ゲートウェイなどが該当する。つまり、対象となるプロジェクトでは、サイバーセキュリティへの適合が、系統連系の条件を満たす設備を選定するうえで、不可欠な判断基準となる。

私は、この取り組みに、極めて実務的な国家戦略の一つの形を見ている。公開された技術基準は、購入者に製品を比較するための共通の尺度を与え、製造事業者には基準への適合を文書で示すことを求める。その商業的な影響は明確である。今後、設備の選定、公的支援の利用、そして電力系統への接続は、こうしたセキュリティ基準への適合状況によって、ますます左右されることになる。

さらに、この適合ラベルは、購入者が次のような重要事項を確認するための出発点にもなる。

• 遠隔アクセス権限を持つのは誰か。

• ソフトウェアの更新は、どのような承認手続きを経て行われるのか。

• 評価対象として登録されたシステム構成には、どの機器が含まれるのか。

• 製造事業者は、どの程度の期間、セキュリティに関するサポートを提供するのか。

電池セルが蓄えるのはエネルギーである。しかし、設備の所有者がそのエネルギーを必要なときに安心して利用できるかどうかは、これらの問いに対する答えによって決まる。

私が2018年にプライベートキーを守るスマートフォンの開発から学んだこと

私は2018年、HTCでブロックチェーン対応スマートフォン「Exodus」の開発に携わった際、これと本質的に共通する課題に直面した。

当時の設計で中心となったのは、利用者の暗号資産を管理する「プライベートキー」を、いかに安全に保護するかという問題だった。私たちは、ハードウェアによって保護された「信頼できる実行環境（TEE：Trusted Execution Environment）」を採用し、通常のAndroid環境からプライベートキーを隔離した。これにより、取引を要求するアプリケーションにプライベートキーを公開することなく、取引への電子署名を実行できる仕組みを構築した。

当時、そこまで徹底した対策を講じる必要があるのかと尋ねられることがあった。しかし、その理由は明確だった。「プライベートキーを使用できる者は、他人が所有する資産に対して、実質的な支配力を持つことができる」

この経験から得た教訓は、今も私の中に深く根付いている。セキュリティの本質は、権限が誰に帰属しているのか、どのような条件でその権限を行使できるのか、そして時間の経過とともに権限がどのように変化するのかという点にある。ハードウェアによる隔離は、その解決策の一部にすぎない。ファームウェア、ソフトウェアの更新手順、そして機器の管理を許可された担当者の存在も、同じように重要である。

これを蓄電システムに置き換えると、問題はより実務的なものになる。

• 誰が充放電の指令を送信できるのか。

• 誰が運転上の制限値を変更できるのか。 ・誰が新しいファームウェアをインストールできるのか。

• 外部サービス事業者との契約が終了した際、設備の運用者は、その事業者に付与していたアクセス権限を確実に取り消すことができるのか。

暗号資産の世界では、プライベートキーに対する管理権限を失うことは、金融資産を失うことにつながる。

一方、蓄電システムにおいて制御権限を失うことは、設備の運用管理や物理的な安全性、さらには停電時に事業を継続するためのバックアップ機能を危険にさらすことを意味する。

したがって、「キーを握っているのは誰か」という問いは、製品が工場から出荷される時点だけで完結するものではない。出荷後に配信されるソフトウェア更新、継続して有効な状態に置かれている保守用アカウント、遠隔サービスに付与されたアクセス権限なども、その対象となる。電力系統の運用者が蓄電システムを接続するとき、受け入れているのは設備そのものだけではない。その設備に付随する、こうしたさまざまな信頼関係も同時に受け入れているのである。だからこそ、これらの関係を理解することは、蓄電池を購入する際の重要な検討事項でなければならない。

現在、私は2018年に直面したものと本質的に同じ問題に、より大きな規模で取り組んでいる。

それは、機器を制御する権限を、正当にその権限を持つべき人々の手に確実に保持するこ とである。

なぜCold Electricにとって重要なのか

Cold Electric（長庚國際能源）は、台湾においてリン酸鉄リチウムイオン（LFP）電池セルおよび蓄電システムの設計・製造を行っている。私たちは、電池セルからモジュール、システムインテグレーションに至るまで、一貫した技術的能力を構築してきた。それは、エネルギー設備に対する責任は、設計、製造、運用の全過程を通じて果たされるべきだと、私が考えているからである。

ユアサM&B株式会社とのパートナーシップによって、私たちはこれらの技術と製品を日本のお客様に提供している。そして今回のJC-STAR ★1適合ラベルの取得は、私たちが今後さらに発展させていくための、客観的に確認できるセキュリティ基盤となった。

私たちは、この適合ラベルを単なるマーケティング上の訴求材料とは考えていない。むしろ、徹底した透明性を実現するための枠組みとして位置付けている。重要なのは、その具体的な内容である。今回登録されたシステム構成には、電池管理システム（BMS）、エネルギー管理システム（EMS）、および現地設備の制御を担うサイトコントローラーが含まれている。これらについて、アクセス制御、ソフトウェア更新、機密情報の保護、脆弱性への対応、製品サポート情報の提供などを含む、16項目の適合基準に基づく評価を実施した。

お客様には、適合ラベルを取得したという成果だけでなく、どの範囲が評価され、どのようなセキュリティ対策によって保護されているのかについても、同じように明確に理解していただく必要がある。私がJC-STAR適合ラベルを高く評価しているのは、私たちの取り組みを可視化し、お客様に具体的な確認・検証の手掛かりを提供できるからである。しかし、その価値は、取得後も継続的な取り組みが行われてこそ維持される。私たちが描く今後のロードマップには、管理体制のさらなる強化、ソースコードおよびソフトウェアのバージョン管理、ファームウェアの検証、製品ライフサイクル全体を通じた一貫した文書管理、そして正式なセキュリティインシデント対応チームの整備が含まれている。製品が進化し、新たな脆弱性が発見される中で、こうした取り組みを継続していかなければならない。

セキュリティとは、製品に備わった固定的な性能ではない。企業が継続して実践し、維持し続けるべき取り組みなのである。

私にとって、ここにこそ「エネルギー主権（Energy Sovereignty）」という概念が具体的な意味を持つ瞬間がある。設備の運用者は、誰が自らの設備にアクセスできるのか、その人物や組織にどのような操作が許されているのか、そして必要に応じてそのアクセス権限をどのように取り消せるのかを把握しているべきである。また、供給事業者は、製品をどのように保守し、問題が発生した際にどのように対応するのかを、明確に説明できなければならない。台湾が持つ製造技術の強みと、ユアサM&Bとのパートナーシップは、私たちがその責任を果たすための重要な基盤となっている。そして、その責任を果たしていることを、実際に納入するシステムの品質と、その後に提供するサポートを通じて証明し続けなければならない。

真の防御とは何か

私は以前、蓄電池がもたらす本質的な価値は「時間を確保すること」にあると述べた。（関連記事：『Security, Time, and Trust』）蓄電池は、病院に対して医療を継続するための時間を与える。

工場に対しては、生産の中断に適切に対処するための時間を与える。そして地域社会に対しては、緊急事態に対応するための時間を与える。

しかし、この価値が成立するためには、一つの重要な前提条件がある。それは、蓄えられたエネルギーに対する制御権を、正当な利用者自身が持っていることである。もし第三者が勝手に蓄電池のエネルギーを使い切ったり、設備へのアクセスを無効にしたり、必要なときに電力を供給できないようにしたりできるのであれば、蓄えられたエネルギーの価値は大きく損なわれてしまう。今後も、価格や電池の化学組成がエネルギー分野における重要な議論の対象であり続けることは間違いない。しかし、それと同じくらい、設備を制御するプログラムや、そのプログラムに指令を与える権限のあり方も、厳格な検証の対象とされるべきである。

日本は今、こうした検証を市場参入の条件として制度に組み込み始めている。台湾も、この動きを注意深く見つめなければならない。台湾新竹市の半導体製造工場に代表される重要産業インフラを守る設備の購入者として、また、国際的な電力インフラ市場での競争を目指す高度な製造技術を持つ輸出国として、台湾には双方の立場から取り組むべき課題がある。明確なサイバーセキュリティ基準を整備することは、国内インフラの強靱性を確保するだけでなく、台湾の製造事業者が国際市場で競争力を維持するうえでも重要となる。

私たちは、自社の蓄電システムがJC-STAR ★1適合ラベルを取得したことを誇りに思っている。しかし、本当に重要な仕事は、ここから始まる。それは、お客様が私たちの製品を信頼し続けられるよう、その信頼に値する理由を、これからも積み重ねていくことである。

蓄電池が本当の意味で時間を生み出せるのは、いつ放電するかを決める権限が、あなた自身の手にあるときだけなのだ。

Subscribe now

English Version:

Now picture the same thing with a battery the size of a shipping container, sitting behind a hospital.

Japan pictured it. Then it did something almost nobody noticed.

Every Battery is a Computer Now

A label is not much to look at: a star, a QR code, a registration number. It is small enough to overlook, yet on August 4, 2026, when Cold Electric’s Taiwan-made BESS series received the JC STAR ★1 conformance label under Japan’s IoT cybersecurity scheme, it represented a pivotal shift. For us, this conformance label was far more than a regulatory milestone; it marked our entry into a market increasingly concerned with who can access, update, and control its energy equipment.

I want to explain why that small label matters. It brings a question into the battery business that belongs beside chemistry, capacity, and cost: whose instructions will this equipment follow?

The Battery as a Computer

A modern battery energy storage system, or BESS, combines electrochemistry with networked control. Three systems are central to its operation.

1. The battery management system (BMS) monitors cell voltages and temperatures and enforces operating limits.

2. The power conversion system (PCS) converts electricity between the battery’s direct current (DC) and the grid’s alternating current (AC).

3. The energy management system (EMS) schedules charging and discharging, sometimes coordinating with an aggregator that manages multiple energy resources, or another remote service.

Firmware governs how these systems behave. Telemetry reports their operating condition, while commands determine when electricity moves.

Recent global security events demonstrate why this control layer requires scrutiny, centered on supply chain trust and ensuring hardware delivers only what is promised. The pager attacks in Lebanon in September 2024 showed how devices people depended on could be turned against them through a compromised supply chain. While those attacks involved planted explosives, a fundamentally different mechanism from cyber risks facing connected batteries, they highlighted the same core concern: how do we know the equipment delivered to us does only what we expect it to do?

That question became concrete for energy in May 2025, when Reuters reported that experts examining some Chinese-made solar inverters and batteries had found undocumented communication devices. Buyers urgently need to know every route through which their equipment can communicate.

For a networked battery, that scrutiny extends to its software, remote connections, and updates. An attacker who compromises the controls could interfere with when it charges, discharges, or remains available. Even without a wider blackout, malicious commands could increase operating costs or drain energy reserved for an emergency. If safety protections are also compromised, equipment damage becomes a concern, meaning a business could lose the backup capacity it paid for precisely when it needs it.

Across a sufficiently large fleet, coordinated changes in charging, discharging, or connection status could threaten grid stability, which is the balance between electricity supply and demand that keeps power reliable. The consequences could reach a hospital forced onto emergency power, a production line in Hsinchu brought to a stop, or a household losing refrigeration and cooling. Their severity would depend on the scale of the attack, grid conditions, and the protections in place.

This is why governments care. Electricity supports water supplies, communications, healthcare, and industry. A vulnerability shared across many storage systems can create risks far beyond the companies that own them. I want a more renewable, decentralized grid, and building it means taking responsibility for the connections that make it work, and for the people who depend on it.

What Japan Did, and How Quietly It Did It

Japan’s approach is to make cybersecurity a central part of the decision to buy and connect equipment.

JC-STAR is an IoT security labeling scheme operated by IPA. Applications for its ★1 level opened on March 25, 2025. At this level, manufacturers assess their products against defined criteria and submit a self-declaration of conformity, after which IPA issues the conformance label through the application process.

Japan has been systematically incorporating that baseline into energy policy. Its fiscal 2025 grid-storage subsidy round added requirements concerning the use of JC-STAR ★1-labeled products, leading to December 2025 approvals of approximately ¥36.3 billion in subsidies across 37 projects. JC-STAR requirements also appeared in the storage provisions of the 2025 long-term decarbonization auction.

The next step reaches grid connection itself. Under the announced rollout, covered equipment in high- and extra-high-voltage solar and storage projects will need a JC-STAR ★1 or higher for grid-connection contract applications accepted from April 2027 onward, with low-voltage projects following in October 2027. The relevant milestone is the contract application, distinct from the earlier connection study. The scope reaches covered control and communications equipment using IP communications, including PCS, EMS, BMS, and gateways. For affected projects, cybersecurity becomes part of choosing equipment that can meet the conditions for connection.

I see a practical form of statecraft in this. A published technical baseline gives buyers a common basis for comparing products and makes manufacturers document their conformity. The commercial consequence is concrete: equipment selection, public support, and access to the grid increasingly depend on meeting that baseline.

The label also gives buyers a starting point for further questions: Who has remote access? How are updates authorized? Which components fall within the assessed configuration? How long will the manufacturer provide security support? The cells store the energy, but those answers help determine whether its owner can rely on being able to use it.

What I Learned Building a Phone That Guarded Keys

I encountered a version of this problem at HTC in 2018, when I helped build the Exodus blockchain phone.

The design centered on protecting the private keys that controlled a user’s crypto assets. We used a hardware-backed trusted execution environment isolated from the ordinary Android environment, so transactions could be signed without exposing those keys to the apps requesting them.

People asked why we went to so much trouble. The reason was clear: whoever could use those keys could exercise power over something that belonged to someone else.

That lesson has stayed with me. Security depends on where authority resides, what can invoke it, and how that authority changes over time. Hardware isolation is one part of the answer, while the firmware, update process, and people authorized to administer the device are part of it too.

In a battery system, the questions become operational: Who can send a dispatch command? Who can change an operating limit? Who can install new firmware? Can the operator withdraw a service provider’s access when the relationship ends? In crypto, losing key authority means losing financial assets; in energy storage, losing command authority risks operational control, physical safety, and backup resilience when the grid goes dark.

“Who holds the keys?” is therefore a question that reaches beyond the moment a product leaves the factory. It includes the software update delivered later, the maintenance account kept active, and the permissions granted to a remote service. When a grid operator connects a storage system, it is accepting those relationships alongside the equipment. Understanding them should be part of buying the battery.

Today, I am working on the same underlying problem at a different scale: making sure authority over the machine remains with the people entitled to exercise it.

Why This Matters for Cold Electric

Cold Electric designs and manufactures LFP battery cells and storage systems in Taiwan. We have built capabilities across cells, modules, and system integration because I believe responsibility for energy equipment should run through its design, manufacture, and operation. Our partnership with Yuasa M&B brings that work to Japanese customers, and receiving the JC-STAR ★1 conformance label gives us a documented security foundation to build on.

Rather than viewing the label as a marketing tool, we see it as a framework for absolute transparency. The details matter. Our registered configuration covers the battery management system, energy management system, and site controller, which we assessed against sixteen conformity criteria spanning access control, software updates, protection of sensitive information, vulnerability handling, and product-support information.

Customers should be able to see those boundaries as clearly as the achievement itself. I value the JC STAR label because it makes our commitments more visible and gives customers something specific to examine. However, its value continues only if the work behind it continues. Our roadmap includes stronger management, source-code and version governance, firmware verification, consistent lifecycle documentation, and a formal incident-response team. That progress must continue as products change and new vulnerabilities emerge.

Security is not something a product has; it is something a company keeps doing.

For me, this is where energy sovereignty becomes concrete. An operator should know who can reach its equipment, which actions they can take, and how access can be withdrawn. A supplier should be able to explain how it maintains the product and responds when something goes wrong. Taiwanese manufacturing and our partnership with Yuasa M&B give us a foundation for taking on that responsibility, and we have to demonstrate it through the systems we deliver and the support we provide.

The Real Shield

I have argued that batteries buy time. They can give a hospital time to sustain care, a factory time to manage an interruption, and a community time to respond.

That promise depends on control. Stored energy is less useful if someone else can drain it, disable access to it, or prevent it from being delivered when needed. Price and chemistry will remain central to the energy debate, but the code commanding the equipment, along with the authority behind that code, deserves the same scrutiny.

Japan is beginning to make that scrutiny part of market access. Taiwan must pay close attention, both as a buyer guarding critical industrial infrastructure like Hsinchu’s fabs, and as an advanced manufacturing exporter aiming to compete in international grid markets. Establishing clear cybersecurity baselines protects domestic resilience while keeping suppliers competitive globally.

We are proud that our systems have received the JC-STAR ★1 label. The work ahead is to keep giving customers reasons to trust the equipment behind it.

A battery only buys time if you are the one who decides when it discharges.

Subscribe now

I was recently interviewed during a visit to Nikkei’s Tokyo headquarters.

You can read the full feature on Nikkei Asia.

On September 16, 2026, Cold Electric helped bring a disaster-response microgrid online at the Beimen District Office on the Tainan coast. On paper, the installation is modest: roughly 53 kW of solar already on the roof, a new 50 kW / 105 kWh battery system, an energy management system, and a generator. During ordinary operation, it reduces the office’s demand for grid electricity at expensive hours. When the grid fails, it separates from the damaged network (islands) and supplies the disaster-response center and shelter with power for communications, drinking water, and lighting.

The building whose job is to coordinate the rescue no longer has to wait to be rescued.

Our batteries began with laboratory research. Our understanding of what communities needed from them deepened in the aftermath of Typhoon Danas. When the storm knocked out power across southern Taiwan in July 2025, we prepared six emergency battery units and sent them into affected communities, including Xuejia, Madou, and Baihe. Each held 90 kWh of electricity and provided 110 V outlets. At local temples, people could charge phones and mobility devices while repairs continued.1

A charged phone meant being able to reach family and receive information. An operating mobility device meant retaining some independence when help was already difficult to reach. The value of the equipment lay in what it allowed people to keep doing.

Beimen carries that lesson into permanent infrastructure. Tainan has announced three larger disaster-response microgrids in Qigu, Madou, and Nanhua, serving the coast, plain, and mountains.2 The temporary reserve we brought into communities after a storm can become a reserve already waiting when the next one arrives.

That is the question a microgrid answers: how can the places responsible for helping the community remain operational when the power system around them fails?

The answer begins with one building. Its implications reach much further. If the institutions we depend on can produce, store, and manage part of their own electricity, they can take back responsibility for a basic condition of community life. That is why there is nothing small about microgrids.

II. The Great Outsourcing

For much of human history, usable energy came from nearby local sources. You cut wood, fed animals, or built a mill where water could turn it. Access was unequal and the work was hard, but the relationship between a community and its sources of power was visible.

Industrialization changed that relationship. Larger engines and centralized electricity generation made it possible to supply more people at lower cost. Power could travel from a distant plant to a factory, a hospital, or a home, without each maintaining its own machinery or fuel supply. The gains were enormous. Modern public health, manufacturing, and everyday comfort grew around them.

We also made a quiet trade. We handed responsibility for the most basic input of modern life to a system beyond our control, in exchange for reliable electricity delivered by someone else. Most days, that dependence is scarcely noticeable. We pay the bill and expect the outlet to work.

A typhoon, earthquake, cyberattack, or blockade exposes the chain behind the outlet. A building may survive while the line supplying it does not. A power plant may remain intact while its fuel deliveries are interrupted. In 2025, imports accounted for roughly 95% of Taiwan’s total energy supply.3 Keeping fuel shipments moving is part of keeping factories, hospitals, and homes running.

The question is how much of that dependence ordinary institutions must continue to accept. A school cannot secure an international shipping route. However, it can prepare to keep a shelter operating when electricity from that wider system stops arriving. Sovereignty begins to acquire a practical meaning when an institution can decide what it will keep working instead of waiting for someone else to restore its supply.

The technology that makes this possible includes a battery energy storage system (BESS), which combines batteries, equipment that converts electricity into a usable form, and controls that manage charging and discharge. It stores electricity from solar panels or the grid so it can be used later. Storage gives an institution a reserve, and with it some choice about when and how to use its power.

III. The Reversal

The solar + storage model changes the practical scale of energy ownership. A school does not need the resources of a utility to install panels and batteries suited to its needs. The equipment can be assembled from repeatable modules, then expanded as demand and funding allow.

Large installations still benefit from economies of scale: their size can spread installation, equipment, and operating costs across more units of electricity or storage. That advantage remains. What has changed is the ability to build a useful system at the place where its electricity will be used. Local control has become an option for institutions that previously had little choice beyond paying a utility bill.

The sun gives us the generation. The battery gives us time.

Without storage, a building must use solar electricity when it is produced or send the surplus elsewhere. The battery lets it keep some of that electricity for the evening, for expensive hours, or for an interruption in supply. It turns generation tied to the sun’s schedule into a reserve the institution can manage.

There is another step. Ordinary grid-connected solar generally shuts down during a blackout to avoid sending power into lines that workers may be repairing. A microgrid needs suitable inverters, protective equipment, and controls to separate safely from the main network and coordinate its own supply. Those controls decide how the solar, batteries, and generator work together and which equipment receives power.

The result is a building or district that can continue designated operations on its own. It remains part of the wider grid during normal conditions. It can reduce its demand at busy hours, store solar electricity for later, and, where local rules permit, supply power or other services to the grid.

A backup battery is useful during an outage. A microgrid also gives its owner a continuing role in producing and managing electricity. The institution begins to understand its own demand, decide which activities matter most, and prepare for the circumstances in which it must rely on itself.

IV. Every School, Every Hospital, Every Church

Where should we begin? With the places people already turn to for help: schools, hospitals, churches and temples, town halls and fire stations.

The small temples where we deployed batteries after Danas became charging points because they were places communities could gather. A school gym can become a shelter. A church kitchen can prepare meals. A district office coordinates the response. They are the civic nodes a society falls back on when everything else fails.

But a building’s role in the community does not guarantee its ability to perform that role during a disaster. A shelter needs light, communications, and drinking water. Refrigeration may be needed for food or medicine. A hospital needs power for patient care. The loss of electricity can interrupt the work people came there to depend on.

When Taipower’s chairman and I discussed microgrids earlier this year, the idea we kept returning to was simple: every school a solar-plus-battery refuge. A school already serves a neighborhood and brings people together around a public purpose. Preparing it to remain useful during a disaster extends that purpose into the hours when the neighborhood needs it most.

The same principle is already taking shape in our work at Cold Electric. A hospital in Kawasaki, Japan, and an Episcopal church in Houston have incorporated our batteries into their power systems. Neither is a power company. Both are taking responsibility for a portion of the electricity on which their work depends.

This requires judgment about each site. A staffed temple with a community hall and an agreed shelter function may need substantial backup power. A small, unstaffed shrine may need only modest emergency equipment. Stewardship carries those obligations. A church that teaches care for creation and a school that teaches civic responsibility have reason to consider how their resources can serve their neighbors. Owning generation and storage gives them a practical way to act on these teachings.

Multiply that responsibility across the places where people learn, heal, worship, and govern, and we begin to build a grid made of refuges. Each institution holds a reserve for its own work and, where it can, for the community around it.

V. Energy, Then Compute, Then Intelligence

The question of ownership does not stop with electricity. It reaches the machines and services electricity makes possible.

Much of our access to advanced AI arrives through distant data centers. A school that wants AI-powered tutoring can buy access to a service, sending students’ questions to systems operated elsewhere. The provider governs access, pricing, and the terms under which the tool may be used. The school gains a capability, but it also acquires a dependence.

That is the great outsourcing again, this time for thought.

The alternative is becoming practical for some uses. Open-weight models make their trained parameters available for local deployment under their licenses. Appropriate models can run on local hardware for selected tasks, giving an institution more choice over how a service is configured and how its data is handled. It need not reproduce the largest frontier model to operate a tool that is useful to its own staff or students.

Consider what that could mean for a school. Its teachers could help choose the material a tutoring system draws on and the conditions under which students use it. A hospital could keep selected records and staff-support tools on local systems. A town hall could maintain access to documents and procedures needed during an emergency. These are specific services an institution can govern, tied to people and responsibilities it already knows.

Integrating batteries and compute into unified containers creates a foundation for local autonomy. A microgrid extends beyond traditional energy infrastructure; it forms the physical foundation for decentralized artificial intelligence, offering a resilient alternative to centralized control.

VI. From One to a Million

Our ambition is ten thousand microgrids in Taiwan, a hundred thousand in Japan, and a million around the world. At that scale, their value extends beyond whether an individual building can survive an outage. We begin to change how the electricity system and the institutions it supplies support one another.

The network of possible partners is extensive. Japan’s government reports more than 150,000 organizations in its shrine and temple categories alone.4 That gives a sense of how widely local institutions are distributed. Japan has also encouraged municipalities to consider disaster-support agreements with religious facilities. The practical work is to identify where those institutions can serve their communities, then design the power needed to sustain that work.

The electrical scale can be illustrated without pretending that every site is alike. Suppose ten thousand installations each have batteries capable of delivering 200 kW and storing 500 kWh. Their combined ratings would be 2 GW of power and 5 GWh of energy. A hundred thousand such installations would total 20 GW and 50 GWh. These figures describe a reserve that must be charged and managed, with its value determined by when and where it can supply electricity.5

On ordinary days, distributed batteries can help with a recurring problem: solar production and electricity demand do not always arrive together. Batteries can charge when solar electricity is plentiful and supply power later, reducing the amount buildings need from the grid after sunset. Across many sites, that coordination can ease pressure on the wider system.

This is the purpose of a virtual power plant, or VPP.6 Software coordinates batteries and other resources so they respond together to the grid’s needs. A school battery might supply power during a busy hour or reduce the school’s demand for electricity from the utility. Under suitable rules and contracts, the operator can receive payment for providing that flexibility. Bill savings and service payments can help finance equipment that also has a community purpose.

The same local reserves change what happens when part of the wider system fails. A damaged substation, transmission corridor, or fuel terminal can affect many places at once. If the institutions depending on that infrastructure have no independent supply, the disruption can interrupt treatment, shelter operations, and disaster coordination across a region. Local generation and storage allow those institutions to keep necessary services operating while the wider system is repaired.

For Taiwan, the value is the time those services buy for the people relying on them. During an interruption in imported fuel, local generation and stored electricity can help sustain essential operations wherever the equipment survives and conditions allow it to work. A hospital that can continue treatment gives patients and responders more time. A shelter with working communications helps people organize, while a district office can coordinate recovery instead of becoming another place waiting for assistance. How long that window remains open depends on the available generation, the battery reserve, and the demands placed on them. That promise rests on equipment people can depend on.

As we build more microgrids, we also take on responsibility for keeping them operational throughout their useful lives. A million microgrids require a sustained supply of battery modules, inverters, controllers, and replacement parts, backed by people who can service them. Communities need to know who built their systems, how they are controlled, and whether their supply chains can be audited.

We want that infrastructure to be built by trusted hands in Taiwan, Japan, and America. Local ownership gives an institution control over part of its power supply; reliable manufacturing and maintenance help it retain that control when circumstances become difficult. Energy sovereignty has to include the ability to maintain what we own.

VII. Owning the Means

The twentieth century fought bitterly over who should own the means of production. For most people buying electricity, the means of energy production remained someone else’s responsibility, somewhere far away.

That arrangement is no longer the only practical choice. A school can own part of its generation. A hospital can manage a reserve. A parish can own its piece of the sun. Where local computing serves their work, those institutions can also take greater responsibility for the tools through which knowledge is stored and used.

Stewardship gives ownership a purpose. A school must decide how it will serve its neighborhood during an outage. A church must decide how its resources can help the people who come to it. Owning a battery brings the obligation to maintain it and the authority to decide whose needs it will support.

There is a larger change in that act. Institutions accustomed to receiving electricity become participants in producing, managing, and sharing it. Their neighbors gain places that can remain useful during a crisis. The wider grid gains resources it can coordinate during ordinary operation. Responsibility is distributed along with the power.

The Tainan microgrid serves one district. It is a start.

Nothing about where it leads is small.

Subscribe now

The theory has a flaw that nobody in Taipei likes to say out loud. It assumes the fabs are running. Every one of them runs on electricity. Roughly half of that electricity comes from natural gas that arrives by ship, and at any given moment the island is holding about two weeks of it. A shield that can be unplugged is not a shield. It is a hostage.

So Taiwan should not fear only the invasion that ends in rubble. It should fear the quarantine that never fires a shot. A few weeks of turned-back tankers, and the most important factories on earth go dark on their own, without a single missile being launched.

I build batteries for a living. My company, Cold Electric, makes energy storage systems, and we have just switched on our first microgrid in Tainan. So it may surprise you to hear me argue that batteries are not the answer to Taiwan’s energy problem. They are part of the answer, and an important part. But Taiwan’s trouble comes precisely from asking each piece of its energy system to do a job that belongs to another piece.

Batteries buy time. Fuel and nuclear buy security. Price buys truth.

Taiwan is in trouble because it has confused the three.

Three problems, not one

When people say Taiwan has “an energy problem,” they are usually describing three different problems that happen to share a single power bill.

1. The first is security. Does the island have enough energy to survive a disruption it does not control? This is a question about fuel: where it comes from, how much of it sits on Taiwanese soil, and how long it lasts.

2. The second is time. Can the grid match supply to demand minute by minute, hold its frequency when a plant trips, ride out a typhoon, and absorb a midday flood of solar without collapsing in the evening? This is a question about the shape of the grid, not the quantity of energy inside it.

3. The third is truth. Does anyone in Taiwan actually know what electricity costs? This is a question about price, and it turns out to be the one that quietly decides whether the other two can ever be solved.

Confuse the three, and every remedy lands on the wrong problem.

I. Security: what fuel and nuclear buy

In the book of Genesis, Pharaoh dreams of seven fat cows grazing by the Nile, only for seven gaunt, starving cows to rise up and devour them. Joseph reads the dream as a warning: seven years of rich harvests will be followed by seven years of famine. His advice to Pharaoh is not to pray for better harvests. It is to build granaries. Store a fifth of the grain during the years of plenty, so that when the famine arrives the nation lives through it. Security is not a mood or an intention. It is inventory.

Taiwan's granary is nearly empty. Imports supply more than 97% of the island's energy. Gas has quietly become the backbone of the grid: it generated 47.8% of Taiwan's electricity in 2025, and gas now makes up more than half of installed generating capacity, up from 36% in 2016. Against that dependence, LNG (liquefied natural gas) stocks sit at roughly 12 to 14 days, barely above the legal floor of 11, and that thin two-week margin has hardly moved in a decade even as import capacity doubled. The supply is dangerously concentrated as well. Qatar alone accounted for 33.7% of Taiwan's LNG imports in 2025, and Beijing has spent years cultivating Doha while rehearsing the encirclement of Taiwan's key ports.

LNG is a superb fuel for a peaceful world and a terrible one for a besieged island. It is bulky, it has to be kept cryogenically cold, and it cannot be stored cheaply for long. You cannot build Joseph’s granary out of liquefied methane.

Nuclear fuel is the mirror image. A reactor’s fuel is so energy-dense that a year or more of it can sit on-site in a space the size of a warehouse. It is the only large, firm source of electricity that Taiwan can stockpile on the scale of years rather than days. That single property makes it the most valuable security asset available to the island, and Taiwan switched it off. At its height, nuclear supplied close to a fifth of the island’s electricity. The last reactor shut down in May 2025.

The encouraging news is that public opinion has already reversed, faster than the law has caught up to it. In last August’s referendum on restarting the Maanshan plant, roughly three-quarters of those who voted said yes. The measure failed only on a technicality, because the yes votes came in just under the quarter-of-the-electorate threshold the law requires to pass. And the machinery of reversal is now turning. A May 2025 amendment allows reactors with expired licences to apply for renewals of up to 20 years, Taipower has filed a restart plan for Maanshan, and the president has come out in favor of bringing both Maanshan and Kuosheng back, subject to safety inspection and regulatory approval. The economics ministry believes Maanshan could be generating again as early as 2028. Countries do not usually un-abandon a technology this quickly. Taiwan is trying to, because the strategic logic has become too loud to ignore.

This is the single highest-leverage decision on the table, because it is the only one that fixes the quantity of energy and the island’s exposure to the sea lanes at the same time. It deserves the urgency of a defense procurement, not the pace of an environmental review.

But 2028 is two winters away, and nuclear alone does not finish the job. The rest of the security agenda is unglamorous, which is probably why it goes undone. Diversify LNG away from any single dominant supplier: the 2025 pledge to buy nearly a third of Taiwan’s LNG from the United States is the right instinct, and it should be the floor rather than the ceiling. Push storage well past the 11-day legal minimum, and treat that buffer as military infrastructure, because that is exactly what it is. And be honest about coal. The island holds months of coal against only weeks of LNG. Coal is dirty and unfashionable, and it is also the deepest granary Taiwan currently has. Its plants should be held in strategic reserve until the reactors return, not retired early for the sake of a cleaner spreadsheet.

None of this is exciting. Security rarely is.

II. Time: what batteries buy

Now to my own industry, and to what it can and cannot honestly do.

A battery does not make energy. It moves energy from one moment to another, and loses a little in the process. Most grid batteries hold two to four hours, which makes storage a timing technology. It is extraordinarily good at the problem of time and completely useless against the problem of security. In a blockade, a battery with nothing behind it lasts an afternoon, and then it is a metal box.

But time is exactly where Taiwan’s grid is quietly coming apart, and that makes the afternoon a battery buys worth far more than it first sounds.

Start with the fastest problem. When the reactors went offline, the grid lost more than megawatts. It lost the enormous spinning mass of their turbines, the physical inertia that used to resist any sudden change in frequency. A grid with less spinning mass swings harder and faster the instant something trips. Batteries answer in milliseconds, faster than any turbine can, holding frequency steady through the first seconds of a fault before another plant can even begin to react. An island grid has no neighbor to borrow stability from. It has to manufacture its own, and increasingly it manufactures that stability out of batteries.

Then there is the shape of the grid itself. It runs as a long north-south spine, with most of the new generation in the south and much of the demand in the north, and twice in recent years a single fault at one southern plant has cascaded into an island-wide blackout. Geography is the enemy here: a disturbance is free to travel the length of the country before anything stops it. Storage breaks that chain by sitting close to the load it protects, at a substation, at a factory, at a district office, so that a failure hundreds of kilometers away stays hundreds of kilometers away.

I have seen what that is worth up close. When Typhoon Danas tore through Tainan, we trucked six of our battery energy storage containers into the worst-hit districts, Xuejia, Madou, and Baihe, and ran them through the night while the grid was down. 90 kilowatt-hours each, a row of 110-volt outlets. They charged phones, kept flashlights and radios alive, and powered the mobility aids that some residents depend on to move. A battery is an undramatic object right up until the moment the lights go out around it, and then it is the only thing in the room that still functions.

That emergency has since become permanent. Last week we commissioned Taiwan’s first district office-level uninterruptible microgrid, at the Beimen District Office, built as the city government’s answer to the blackouts and communications failures that Danas had exposed. We tied the building’s existing solar array and diesel generator to a new battery, all run by a single management system. On a normal day it shaves the peak and works the time-of-use tariff, quietly lowering the power bill. When the grid goes down, it slips into island mode and carries the building on its own for more than 48 hours, long enough to keep the command center staffed, the shelter lit, the water running, and the phones connected while the emergency is still unfolding. That is what storage near load actually looks like. A public building that stays lit while the district around it goes dark.

The district office has to survive for two days. The most valuable buildings in Taiwan have to survive for a quarter of a second. Inside a semiconductor fab, a voltage sag too brief for a household to notice, a dip lasting a few hundred milliseconds, can wreck a production run and scrap wafers worth millions. No power plant, however large, can act that fast. By the time a turbine has even registered the disturbance, the wafers are already gone. A battery sitting at the fab bridges the gap before it opens. For the industry the entire silicon-shield theory rests on, that quarter of a second is the difference between a normal Tuesday and a 9-figure write-off.

Storage does the slower work too. Wires, transformers, and substations are all sized for the single worst hour of the year, which means most of that copper and steel sits idle almost all the time. Shave the peak and you can postpone years of expensive construction. New York’s Brooklyn-Queens program spent about US$200 million on distributed resources to defer a substation that would have cost a billion.

And storage is what makes Taiwan’s solar build-out usable in the first place. Without it, midday solar floods the grid and then vanishes at sunset, precisely as the evening load begins to climb. With it, the noon surplus becomes the evening supply.

There is one more thing, which I will only gesture at here. A battery on its own cannot survive a blockade. A battery paired with a solar array is a different machine altogether, a generator whose fuel cannot be interdicted, because nobody can blockade the sun. It is the one supply Taiwan can add that a quarantine has no way to reach, and it points toward an energy shield assembled not from a few great plants but from thousands of small ones. That argument deserves its own essay.

So this is what batteries actually buy: seconds of frequency, a quarter-second of protection for a wafer, hours of shaved peak, years of deferred construction, and 48 hours of light in a district office with the grid down. All of it is valuable, fundable, and necessary. None of it is security. Anyone who tells you storage is the answer to Taiwan’s fuel problem is selling you something.

III. Truth: what price buys

The third problem is the one that makes the first two so hard to solve.

In 1945 the economist Friedrich Hayek made a simple but powerful argument: a price is not just a number, it is a piece of information. No government, however capable, can track everything happening across millions of separate households and factories at once, how much each one has, how much it needs, how badly it wants more. A market price does that job automatically. It gathers all of that scattered knowledge and compresses it into a single figure that tells everyone what to do, without anyone giving an order.

Electricity shows this cleanly. When a heat wave pushes demand up on a summer evening and power grows scarce, a real price rises, and it does not need to explain why. A factory sees the higher number and delays the shift that can wait. A household sees it and runs the dryer at midnight instead. Nobody told them to. The price carried the message, and each of them answered it in their own way.

Now fix that price by political decree. The electricity is no less scarce than before, but you have deleted the message. The factory and the household never learn that this was the worst hour to draw power, so they draw it anyway, and the strain simply reappears somewhere else: on the grid, on the utility’s balance sheet, in the next blackout.

Taiwan has done exactly this, only not for a single hot evening. It has done it to its entire electricity price, permanently, as a matter of policy. That price is a deleted message. The island has among the lowest industrial and residential power rates in the world. Households pay about NT$2.77 per kilowatt-hour, while Taipower’s cost to generate that same unit has stood near NT$3.8, so every unit sold to a home loses close to a New Taiwan dollar before a single wire is even paid for. The average tariff only looks close to cost because industrial users quietly subsidize the gap. The damage shows up on the balance sheet: Taipower’s accumulated losses reached NT$417.9 billion by mid-2025, and a NT$100 billion rescue package was blocked in the legislature. This month the pattern repeated itself. Taipower asked to raise rates by about 12.83%, and the average rate was frozen instead, held at NT$3.78 until the next review.

A low price feels like a gift. It is actually a loan, and someone always repays it: through Taipower’s balance sheet, through taxpayer bailouts, through a grid that never gets rebuilt. The scarcity does not disappear because the price hides it. It simply gets communicated through a worse channel, later, at higher cost.

The deeper damage is that a false price stops people from making good decisions. When power is sold below what it costs, nobody insulates a building, nobody shifts usage to off-peak hours, nobody invests in efficiency, and no business can rationally decide whether to generate and store its own electricity, because the number it would compare against is fictional. Every investment that might reduce Taiwan’s dependence on imported fuel is made to look uneconomic by the very subsidy meant to protect the public.

This is why price is the hinge on which everything else turns. Get it right, and private capital will finance efficiency, storage, and rooftop generation on its own, without the state spending a dollar. Get it wrong, and every other reform has to be dragged uphill by subsidy and mandate.

Truth, though, does not have to mean cruelty.

The energy ministry has already done impressive work encouraging commercial and industrial users to install Taiwan-made battery cells. The next step, once the real price is restored, is to protect people directly rather than through the meter. Subsidize households, not kilowatt-hours. Move tariffs toward true cost, and use targeted transfers to shield low-income families from the shift. Price electricity by the hour, so that the evening peak costs what it actually costs and the noon surplus is cheap. A household protected through its income, rather than through a falsified price, can see the truth and still pay its bills.

What a serious energy shield looks like

Put together, the program is not complicated. It is only politically hard.

For security, restart Maanshan and Kuosheng with something close to wartime urgency. Diversify gas away from any single supplier, and expand the buffer far beyond eleven days. Hold coal in strategic reserve until the reactors are back.

For time, build storage where it does work that no other asset can: fast frequency response to replace the lost inertia of the turbines, sub-second protection for the fabs, capacity near the northern load so that a southern fault stays local, and firming for every new solar farm. Where solar meets storage on rooftops, schools, and factories, the grid gains thousands of small generators that no blockade can starve.

For truth, let the price speak. Move tariffs toward cost. Protect the vulnerable directly. Price time honestly.

Each of these depends on the others. Nuclear without honest prices becomes just another state-financed deficit. Storage without honest prices has no market signal telling it where to go. And honest prices without security still leave the island two weeks from darkness.

Taiwan spent 40 years building its silicon shield with remarkable discipline: patient capital, long horizons, a willingness to do difficult things before they became fashionable. It has spent the last decade treating energy as a political convenience rather than a strategic asset. It asked renewables to deliver security, which they cannot do alone. It asked subsidies to deliver affordability, which only hides the bill. And it asked no one at all to deliver the truth.

The island needs an energy shield to protect its silicon one. Batteries buy time. Fuel and nuclear buy security. Price buys truth. Taiwan can have all three, but only once it stops asking any one of them to do the work of the others.

Friedrich Nietzsche, the nineteenth-century German philosopher, gave it a name: slave morality.

In On the Genealogy of Morality1, he made the most damaging attack on Christianity from outside the tradition. The weak, unable to defeat the strong, do something cunning: they revalue the values. They rebrand weakness as virtue. Meekness becomes holiness, suffering becomes sacred, turning the other cheek becomes the highest moral act. The losers don’t beat the strong on the field — they slowly colonize the culture with resentment until the field itself is declared immoral.

Christianity, in his reading, is the institutionalization of that resentment. It takes people who could have said yes to life in full — created, risked, willed their vision into existence against resistance — and teaches them instead to be patient, passive, self-diminished. His counter-figure, developed in Thus Spoke Zarathustra2, is the Übermensch — roughly translated as the Overman, the one who has surpassed ordinary human limitation. He creates his own values, wills past what is comfortable, needs no heaven as compensation for failure on earth, and whose gifts are sovereign expressions of a sovereign self.

That critique deserves to land before it gets answered. A Christianity that uses faith primarily as comfort — that tells people to endure because heaven compensates, celebrates passivity as humility, and mistakes the burial of gifts for the holiness of self-denial — that Christianity is exactly what Nietzsche says it is. It produces people who are passive, quietly resentful, and safe. And it is a profound misreading of the tradition it claims to represent.

The place to see why is the Parable of the Talents3.

In the parable, a master preparing to travel entrusts three servants with sums of money — called “talents,” the largest unit of currency in the ancient world — proportional to each servant’s ability. Two servants invest their talents and double them. The third, afraid of losing what he was given, digs a hole and buries his talent in the ground until the master returns. When the master comes back, he praises the two who risked and multiplied. The one who buried his talent — who played it safe and returned the principal untouched — is stripped of what he was given and cast into outer darkness.

The servant who protects himself from all risk, who buries his gift and waits quietly, is the one condemned. The meek preservation of the principal earns fury, not blessing. The master has no patience for the quiet endurance of the buried talent.

This is not slave morality. This is the tradition’s refutation of slave morality from within.

The servants who are praised entered the game. They took risk. They put themselves and what was entrusted to them into conditions where loss was genuinely possible.

So far, Nietzsche and the parable agree: maximum engagement is demanded, passivity is condemned, the gifts must be deployed at full force. The Übermensch reading is a genuine corrective to the slave-morality misreading of faith.

But the parable is not the Übermensch, and the difference runs all the way down.

The Übermensch worships the gifts themselves. His will, capacity, and power are the locus of value. The gifts are his — expressions of a sovereign self. The telos4 — the purpose toward which the action is directed, the end the whole thing is aimed at — is self-determined. External achievement is the proof of will. There is no Master, no direction beyond the individual’s own self-overcoming. The game exists to demonstrate what he is made of.

The parable inverts every one of those claims. The gifts are not yours. You are a steward — an oikonomos, a Greek word for the manager of a household, someone entrusted with resources that belong to another. The telos is given, embedded in the purposes of the Master who will return and ask what you made of what he gave you. And the external goods — the doubled money — are not actually the point.

Consider what the master’s praise reveals: the two servants who doubled their money receive identical praise despite entirely different absolute returns. The one who turned five talents into ten and the one who turned two into four are commended in the same words. The master is not evaluating the size of the outcome. He is evaluating the quality of engagement — whether the servant genuinely entered the practice, took the risk, participated fully. The money is evidence of engagement, not the measure of it.

This is the distinction philosophers draw between external goods and internal goods5. External goods are what you can show for a practice: the money made, the reputation earned, the result you can take away and use elsewhere. Internal goods are what the practice makes of you: the judgment, the character, the capacity that only exists because you went all the way through it. A surgeon’s clinical instinct is an internal good — it only exists because she operated on thousands of patients and cannot be transferred to someone else the way money can. The master praises the two servants not because they doubled the money but because they became, through the practice of faithful risk, the kind of people who could be trusted with more.

The essay’s argument follows from this: faith is not slave morality, and its alternative is not the Übermensch. It calls for full engagement — the Übermensch’s fierce refusal of passivity — without the Übermensch’s self-worship. The gifts are deployed at maximum intensity, but as a steward who answers for them. The internal goods that emerge through that engagement — genuine self-overcoming, capacity forged through difficulty, becoming more than what you were — are what the master is actually after. You do not produce them by pursuing them directly. You produce them by entering the conditions where they can only come from: genuine risk, genuine uncertainty, genuine exposure.

There is one respect in which this is harder than pure Nietzsche. The Übermensch can console himself with the quality of his willing even when external results don’t materialize: the nobility of the attempt is itself the victory. The faithful steward cannot fully rest there. He answers to someone else’s standard. The reckoning is real. You are accountable not only to your own sense of self-overcoming but to the Master who entrusted the talent and will return to account for it.

Faith is not a crutch. Properly understood, it is the most demanding posture available: full engagement with gifts you don’t own, toward a purpose you didn’t design, under genuine uncertainty about outcomes, judged not on the returns alone but on the person you became through the attempt.

What follows is a seven-level argument for why antifragility is the deep structure of this kind of faith — moving from what kind of thing a soul is, to how it knows, to how it endures, to how it acts, to how it looks when actually lived, to why the tradition that carries this knowledge has survived, to what single image holds it all.

Level One: Ontology — What Kind of Thing Are You?

Fragile, Robust, Antifragile

Nassim Nicholas Taleb6 is a philosopher, mathematician, and former options trader who spent his career thinking about risk, uncertainty, and how systems behave under extreme stress. His central observation is simple and devastating: the modern world systematically misunderstands what it means for something to be strong, and the misunderstanding has consequences at every scale from the personal to the civilizational.

English had a word for what breaks under stress — fragile — and a word for what resists it — robust. But no word for what gains from stress. Taleb argued the absence was telling, that the missing category was the most important one, and coined the word: antifragile.

These are not three points on a spectrum. They are three genuinely different relationships to disorder. The fragile breaks: the wine glass dropped on stone. The robust resists and returns to baseline — think of the Phoenix7, the mythological bird that burns to ash and is reborn from those ashes as exactly what it was before. The antifragile gains — think of the Hydra8, the serpentine monster from Greek mythology that grew two heads for every one that was cut off. Every attack made it more dangerous. The same shock that destroys the fragile and merely inconveniences the robust makes the antifragile more.

Most Christian theology as it gets practiced week to week is Phoenix theology. The goal is resilience: survive the hard thing, return to who you were before, get back to baseline. The sermon series is called Bounce Back. The testimony is about restoration. The frame is recovery. This is not wrong — resilience is better than breaking. But it is not the deepest register of the tradition, and it leaves the door open to Nietzsche’s critique: a theology whose primary register is endurance does begin to look like finding comfort in suffering.

The Übermensch, placed in Taleb’s triad, is also Phoenix territory. He is not fragile — he does not break. But he is robust, not antifragile. He wills through difficulty, overcomes, endures, and returns to himself. The self is the fixed point. Identity is preserved and even strengthened, but it remains fundamentally the same entity returning to baseline with greater force. The Übermensch suffers and is reborn as exactly what he was — only more sovereignly so.

This is why the corrective to slave morality cannot stop at the Übermensch. Antifragility requires something the Übermensch framework cannot accommodate: genuine openness to being changed by what you encounter. Not strengthened — changed. Not returning to yourself with greater force, but becoming something you were not, something that could not have been predicted from what you were. The Übermensch creates his own values; he does not receive them from encounter with otherness. He wills, he does not yield. A self that cannot yield cannot be genuinely transformed.

The Resurrection is not Phoenix theology.

Jesus does not die and return to baseline. He enters the conditions of maximum disorder — betrayal, unjust trial, torture, death, burial, sealed stone, Roman guard — and emerges as something that had not existed before. A body that is physical but no longer mortal. A new creation breaking in from the future. He is recognizable but different: he eats fish, walks through walls, bears the wounds without being defined by them.

This is Hydra theology. The worst possible shock produced not a return to the previous state but an entirely new order of being.

The soul is not designed for preservation. It is designed for antifragility — to become, through the specific stressors of a faithful life, something that could not have existed without them. Everything else follows from that.

Level Two: Epistemology — How Do We Know Anything?

Trial and error, optionality, via negativa

If the soul is designed to be transformed by what it encounters, then the method of knowing has to match. You cannot navigate toward transformation from a position of safety. Transformation requires entering conditions where it is possible — which means entering conditions where failure is possible.

God does not give certainty before commission. He gives direction and expects movement. Abraham: “Go to the land I will show you” — not coordinates, a direction. The disciples: “Go and make disciples of all nations” — no playbook, no guaranteed outcome. Jesus to the first followers: “Follow me.” Each call is structured the same way: enough information to take the next step, not enough to guarantee safety.

But the “going” is not merely how Abraham learned. It is what made him. Abraham did not become the father of three monotheistic faiths by deliberating about the land he would be shown. He became it by going. The act of departure was generative: the going created nations, cultures, and belief systems that could not have existed before he took the first step into territory he could not see. He is not remembered for the quality of his analysis. He is remembered for having moved before he had the information that would have justified moving. The movement produced the information, and the movement produced him.

Brian Armstrong, the founder of Coinbase, has a version of this that circulates widely among founders: action produces information. The instinct under uncertainty is to study, model, and wait until the picture is clear enough to justify the move. But the picture does not clarify from a standstill. His image: climbing a mountain in heavy fog — you can see only a few steps ahead, so you take them, and taking them reveals the next few. You cannot see the path from where you are standing. You can only see it by walking it. Ship the product, and the moment it meets the world you often discover you built it wrong — and that discovery, available no other way, is the most valuable thing you now possess. The error is not the failure of the process. The error is the output of the process.

The structure is the same as Abraham’s commission and the disciples’ charge: you do not receive the knowledge and then act. You act, and the action produces the knowledge — knowledge that changes you into someone capable of the next step. Faith is not certainty that justifies movement. Faith is the willingness to move before the justification exists, trusting that the movement itself is how the path, and the self, are made.

Via negativa9, knowing by subtraction, is the epistemology that follows from this. Taleb names it, but the principle is older: you know more clearly what does not work than what does. What God is not than what God is. What you are not capable of than what you are. Each failure subtracts an illusion. What remains after the subtraction is more real than what was there before.

This is the logic of the furnace. Gold does not become gold by the addition of gold. It becomes gold by the removal of everything that is not gold. The fire does not add anything. It takes away.

Optionality is the structural form this epistemology takes: cap the downside, leave the upside open. Never be in a position where a single error is fatal. Taleb calls this the barbell10 — maximum safety at one end, maximum exposure at the other, nothing in the mediocre middle. The barbell is not risk-avoidance. It is risk-architecture: survive any single failure, so the one time you are right it compounds into something that matters.

Paul in prison, writing to the Philippians while awaiting a verdict that might mean death11, is the barbell in action. Maximum security in his identity in Christ; maximum exposure in his physical circumstances. His peace does not depend on whether Rome releases him. It is grounded in a conviction that the Master is trustworthy regardless of outcome. From that foundation, any level of exposure is survivable.

Level Three: Psychology — How Do We Relate to Uncertainty?

Tranquility, the love of errors, hormesis

Trial and error is only possible for someone who can tolerate — more than tolerate, genuinely accept — the possibility of loss. Most people cannot. The reason is not weakness. It is a preference that operates below the surface of conscious decision-making and, left unexamined, functions as a theology.

The fragile craves tranquility.

Not just financial stability, but psychological and existential stability too — the guarantee that tomorrow will resemble today, the elimination of unpredictability from a life. This craving is a theological statement even when it doesn’t know itself to be one. It says: I trust only what I already hold. I am not willing to stake what I have on what I might find. I am not willing to find out what I am made of.

Aleksandr Solzhenitsyn12 — the Russian novelist and dissident who survived Stalin’s labor camps and spent his career documenting the Soviet system’s machinery of oppression — understood this craving at civilizational scale. At Harvard in 1978 he told the West what it did not want to hear: the elimination of suffering from daily life had not produced flourishing. It had produced fragility. Russians who had survived deep suffering had developed a spiritual intensity the comfortable West could no longer recognize or produce. He was describing what ecologists call the wildfire suppression problem: protect a forest from every small fire and you guarantee a catastrophic one. A civilization insulated from all disorder builds up the conditions for a disaster it cannot survive.

The medical name for the therapeutic principle is hormesis13: small doses of a stressor improve resilience. Vaccines work by introducing a weakened pathogen. Exercise stresses muscle until it breaks and rebuilds stronger. The body grows through manageable exposure to what might otherwise harm it. The soul works the same way, and the ancient practices of fasting, voluntary privation, and deliberate self-denial were applying this principle long before the science named it.

The psychological signature of the antifragile soul is not the tolerance of errors. It is love of them. Not performed nonchalance about failure — actual recognition that errors are information, the specific data through which a finite creature navigating an open creation discovers what is true about the world and about herself. The person who cannot love her errors cannot learn from them. She arranges her life to produce no errors at all — and in doing so, arranges her life to produce no discoveries, no self-knowledge, and no development of the capacity to be trusted with more.

Paul, before the Damascus road, had been Saul of Tarsus14: a devout Jewish scholar who considered the early Christian movement a dangerous heresy and had devoted himself to hunting down and imprisoning its members, complicit in the killings of people he would later call brothers. Total error. And in God’s economy, that error became the specific qualification for his apostleship — not despite having been the church’s most violent persecutor, but because of it. He knew the opposition from the inside. He knew what the reversal that grace required actually felt like. He knew, with a precision no one else possessed, what it meant to be wrong about everything that mattered and to discover you were loved anyway. “I am the least of the apostles, unworthy to be called an apostle, because I persecuted the church of God. But by the grace of God I am what I am.” The error became his credential.

Level Four: Ethics — Who Bears the Risk?

Skin in the game, soul in the game

Trial and error creates an ethical problem the previous levels haven’t addressed. When someone takes a risk and it fails, someone bears the cost. The antifragile framework is not neutral about who that should be.

Taleb’s most theological concept is skin in the game15. Those who make decisions must bear the consequences of those decisions. The absence of skin in the game produces fragility in systems: people take risks they do not personally bear, export the cost to others, and extract the upside for themselves while manufacturing harm for everyone around them. The banker whose bank is bailed out. The consultant whose advice fails without personal consequence. The expert who recommends treatments he would not use himself. Each of them has extracted safety at someone else’s expense.

Above skin in the game is what Taleb calls soul in the game. Not just financial exposure, but identity, meaning, and vocation staked on the outcome. Skin is what you can lose. Soul is what you are. The mercenary has skin in the game. The missionary has soul in the game. The bureaucrat has neither.

The Incarnation is God putting soul in the game at full exposure.

He does not observe the human condition from safety and issue directives. He enters it — fully, physically, irreversibly. Hunger, exhaustion, temptation, rejection, grief, the terror in Gethsemane, the specific desolation of the cry from the cross: “My God, my God, why have you forsaken me?” Taleb, not himself a Christian, names this exactly: “Christianity, by insisting on the Trinity, managed to allow God to suffer like a human, and suffer the worst fate any human can suffer. Thanks to the complicated consubstantial relation between father and son, suffering was not a computer simulation to the Lord but the real, real thing.”16 Soul in the game. Not observation. Participation.

Taleb’s ethical commandment follows: thou shalt not have antifragility at the expense of the fragility of others. The person who imports safety for himself while exporting disorder to those he was meant to serve is the ethical villain. Taleb calls him the fragilista17 — the one who makes systems fragile while believing he is making them safer. The helicopter parent who produces a fragile adult. The comfortable preacher who never asks anything demanding of his congregation. The institution that protects itself from disorder and passes the cost to the people who depended on it.

The third servant is a fragilista in miniature. By burying the talent he protected himself from any risk of loss. But stewardship is not a private arrangement. A talent deployed in the world generates downstream effects: commerce conducted, capacities developed, models of faithful risk that others can see and follow. When he buried his talent, he withheld all of that. He deprived the master of a return, the community of whatever value the talent in motion would have created, and himself of the only conditions under which he could have become someone worth trusting with more. He kept the principal safe and impoverished everything.

Level Five: Biography — How Do Antifragile Lives Work?

The stressor as forge

The previous four levels are structural claims. This one is evidentiary. The question is whether antifragile formation actually appears in human lives — whether real people in conditions of genuine adversity emerge as something constitutively new rather than merely damaged or restored.

The pattern, across every such life, is the same: the stressor was not incidental to the calling. It was load-bearing. Remove any single descent and the architecture of what they became collapses.

Joseph.

In Genesis18, every stressor is structural. Thrown into a pit by his brothers. Sold into slavery for twenty pieces of silver. Falsely accused by Potiphar’s wife. Thrown into prison. Forgotten there for years. Each descent seems more final than the last. Then: he interprets Pharaoh’s dream and becomes second-in-command of Egypt, administering the grain reserves that keep the ancient world alive through seven years of famine — including the brothers who sold him, including the father who thought him dead.

The sentence he says to his brothers is the most economical statement of antifragile biography in scripture: “You meant evil against me, but God meant it for good.”19 Not: despite the evil, good came. The evil was the mechanism. The pit was not an obstacle on the road to the palace — the pit was the road. Without the pit, no slavery. Without the false accusation, no prison. Without prison, no proximity to Pharaoh’s cupbearer. Without the cupbearer, no dream. Remove any link and the architecture of salvation dissolves.

Dostoevsky.

December 1849, Semyonovsky Square, St. Petersburg. Fyodor Dostoevsky20 — a twenty-eight-year-old writer arrested for involvement in a utopian socialist discussion circle21 — stands sixth in line for execution. He watches the first three prisoners tied to posts and blindfolded. By his own calculation he has approximately one minute left to live. The reprieve arrives at the last possible moment. One prisoner, Nikolai Grigoriev, goes permanently insane on the spot. Dostoevsky writes to his brother that night: “Life is a gift, life is happiness, each minute could be an eternity of bliss.”

Then four years of hard labor in Siberia. Ten-pound leg irons. One book permitted — the New Testament given to him by the wives of the Decembrists, Russian aristocrats who had led a failed uprising against the Tsar in 1825 and been exiled to Siberia as punishment, whose wives had followed them voluntarily and for decades distributed scripture to arriving prisoners. Before Siberia: a promising minor writer with secular progressive sympathies. After Siberia: Crime and Punishment, The Idiot, Demons, The Brothers Karamazov — every masterpiece. He chose John 12:24 as the epigraph for Karamazov and as the inscription on his own tombstone: “Unless a grain of wheat falls into the earth and dies, it remains alone; but if it dies, it bears much fruit.” Near the end of his life he wrote: “My hosanna has come forth from the crucible of doubt.”22 The crucible was not incidental to the hosanna. It was the instrument.

Solzhenitsyn.

Arrested in 1945 for criticizing Stalin in private letters. Eight years in Soviet labor camps. With neither pen nor paper permitted for serious composition, he memorized entire novels in his head, line by line, using a rosary made of bread pellets to count his place in the manuscript he was carrying in memory alone. The Gulag Archipelago — the most comprehensive account of the Soviet terror ever written — could only have been produced by someone who had lived inside it. The stressor was constitutive of the work. From the camps came one of the great sentences of the twentieth century: “Bless you, prison, for having been in my life.”23 Not gratitude as performance — gratitude as a precise accounting: without prison, no book; without the book, no exposure of the system that the book eventually helped to bring down.

Jensen Huang.

At smaller stakes and in a different domain, the same structure appears. The near-bankruptcy of Nvidia in 1996, thirty days from going out of business, forced a complete abandonment of its chip architecture mid-production. Subsequent bets that nearly destroyed the company. The 2006 pivot to CUDA24 — a programming framework allowing graphics chips to handle general computation — made when no market existed to validate it. Each crisis clarified what was real and eliminated what was not. He told a Stanford audience: "I wish upon you ample doses of pain and suffering." Not brutality. Accuracy. The Nvidia that now underpins the infrastructure of the AI economy is not what survived those near-deaths. It is what they produced.

Across all four: the stressor reveals what is real. Conviction that cannot survive disorder was not conviction. Faith that cannot survive the furnace was not faith. The stressor does not add something new — it removes the illusions that were concealing what was already there.

Level Six: Civilization — The Lindy Effect

What survives, and why

The previous level showed what antifragile formation looks like in individual lives. The question now is whether the practices that produce it are themselves antifragile — whether the tradition carrying this knowledge has survived precisely because it encodes something structurally true.

Taleb’s Lindy effect25: things that have survived a long time are more likely to keep surviving. The name comes from Lindy’s, a New York deli where Broadway comedians observed, informally, that a comedian who had been performing for twenty years would likely perform for twenty more, while a newcomer might vanish next season. Taleb formalized the principle: for non-perishable things like ideas, institutions, and practices, longevity is evidence of antifragility. Whatever has survived repeated disorder has already proven it can withstand it.

The stories in this essay are old for a reason. Joseph’s pit and palace have been read for three thousand years because the pattern they describe is not culturally specific — it is the deep grammar of how human beings are formed under pressure. Paul’s reversal has held across every culture, language, and circumstance it has encountered because the structure of catastrophic error becoming qualification is not limited to the first-century Mediterranean. Dostoevsky’s prison produced novels read in languages he never knew because what he was describing about suffering and transformation is structurally true, not accidentally so.

The ancient Christian practices of fasting, self-denial, and voluntary exposure to difficulty — practices that much of Western Christianity has largely abandoned in favor of comfort and relevance — were building antifragile souls through deliberate hormesis. Small doses of voluntary stress, chosen and sanctified, producing depth that protection cannot. The tradition knew what it was doing long before it could articulate it in Taleb’s vocabulary. Taleb is the translator, not the inventor.

A Christianity that promises comfort and the removal of difficulty produces fragile souls — optimized for good conditions, underprepared for genuine disorder. When the disorder arrives, as it always does, there is nothing to meet it. The practice of difficulty was not cruelty. It was formation.

Decentralization is the structural form this takes at scale. Antifragile systems are distributed rather than concentrated. The early church grew fastest under persecution because each shock decentralized and spread the movement. Tertullian, the early Christian theologian writing in the second century, put it precisely: “The blood of the martyrs is the seed of the church.”26 The centralized Roman Empire was fragile — concentrated, rigid, unable to adapt to shocks from its periphery. The decentralized network of communities meeting in houses, sharing meals, bearing one another’s burdens, was antifragile: cut off any node and two appeared elsewhere, and every attempt to suppress the network made it more numerous.

Level Seven: The Image — Fire and Wind

Fire and wind are not decorative images in scripture. They are the primary images for the presence and action of God — and the choice is not accidental. Both destroy what is fragile and strengthen what is not. Both are uncontrollable. Both are, depending entirely on what you are when you meet them, either what eliminates you or what defines you.

Fire: the burning bush27 that is not consumed, where God first speaks his name to Moses — present in the form of a thing that should destroy what it inhabits but does not. Malachi’s refiner’s fire28, whose purpose is not destruction but purification. Tongues of fire at Pentecost, resting on each disciple individually, the Spirit embodied in the form of what clarifies by burning. Peter: faith “of greater worth than gold, which perishes even though refined by fire.”

The forge is not God’s punishment. It is God’s presence. The fire that burns away the dross is not a different fire from the one that purifies what remains. To encounter God is to encounter fire — and what survives the encounter is more itself than what entered.

Wind: the ruach29 — a Hebrew word meaning both wind and spirit, breath and divine presence — hovering over the face of the waters at the beginning of everything. Jesus to Nicodemus: “The wind blows where it wishes, and you hear its sound, but you do not know where it comes from or where it goes. So it is with everyone who is born of the Spirit.” Pentecost: “Suddenly there came from heaven a sound like a mighty rushing wind, and it filled the entire house” — and then the tongues of fire.

The wind cannot be controlled, predicted, or domesticated. It can only be received — or hidden from.

A candle hides from the wind because it has to. A small, self-contained flame knows what wind does to flames like it. The protection is rational, even necessary — but it is also the very thing that keeps the candle small. It will always be a candle. The condition of its survival is the condition of its limitation.

The fire does not hide from the wind. The same force that extinguishes the candle feeds the fire, because the fire has been through enough heat to be constituted by it rather than threatened by it. It does not merely survive the wind. It uses the wind. The disorder is not an obstacle to its growth. The disorder is the mechanism.

The difference between the candle and the fire is not talent or circumstance. It is whether the soul has been through enough fire to become fire — to pass through the furnace willingly enough, and often enough, that the wind no longer threatens it. Only enlarges it.

The Proof of Concept

The cross is the maximum-disorder event in history. God incarnate at full exposure to every stressor: betrayal by a friend, abandonment by those he had formed, unjust trial, public humiliation, torture, death, burial, sealed stone, Roman guard, three days of silence. The ultimate buried talent. The grain of wheat at the furthest point of descent. By every external measure, Good Friday is the permanent end of the story.

Easter is what antifragility looks like at full power.

Not the grain preserved. Not the grain restored to what it was. The grain that entered the ground and became a harvest that could not have existed any other way. “Death is swallowed up in victory”30 — not defeated from without, consumed from within. The ultimate stressor became the mechanism of the ultimate flourishing. The maximum disorder event produced not a return to baseline but a new creation.

This is the ground claim of the Christian faith: the structure of reality is antifragile at its deepest level, and the Resurrection is the demonstration of that structure at maximum scale.

Every antifragile pattern in human experience — Joseph’s pit becoming the palace, Paul’s persecution becoming his apostleship, Dostoevsky’s mock-execution producing the novels, Solzhenitsyn’s Gulag producing the indictment that helped bring down the system, Huang’s near-bankruptcies producing the architecture of a new technological era — is the fractal expression of this. Fractal in the precise sense: the same structure repeating at every scale, from the personal to the civilizational to the cosmic. Not coincidence. Not pattern-matching imposed after the fact. The echo of the deep structure of reality, demonstrated once in a sealed tomb on the third day and reverberating through every faithful life since.

The grain of wheat dies. It multiplies. This is how the world was made to work.

Be the Fire

The wind is coming. Not as a possibility — as a certainty. Disorder, stressors, shocks, errors, the specific difficulties native to a life faithfully lived in an unfinished world: these are not exceptional events. They are the normal conditions of existence for a creature designed for antifragility.

The question is not whether the wind comes. The question is what you are when it arrives.

Everything the tradition asks of you — the trial and error, the love of errors, the willingness to stake skin and soul rather than protect them, the ancient practices that looked like deprivation but were actually formation — none of it is the cost of a faithful life. It is the mechanism of it. The wind is not what you endure on the way to becoming. The wind is how you become.

God’s purpose for your life is not a script to execute in conditions of safety. It is a posture to inhabit in conditions of uncertainty: conviction deployed at cost, in the direction of a future you cannot yet prove, for the sake of something larger than your own preservation.

You were not made to be a candle. You were made to be a fire.

Wish for the wind.

Subscribe now

This essay has described the structure from the outside — seven levels, the biographical portraits, the Resurrection as ground truth, the image of fire and wind. What it has not described is the phenomenology of the encounter from the inside: what it actually feels like to be the grain entering the ground, what happens when you permit the practice to claim you fully, how a person becomes genuinely different rather than merely stronger.

The first essay, The Venture Gospel, argued that the Parable of the Talents is not about returns but about the internal goods of the practice. The third, The Fusion of Horizons, will explore what the encounters of the practice actually do to you — and why time changes everything you see. Each essay stands alone.

“Get rich or die tryin”— 50 Cent

Marc Andreessen1 said it on a podcast: “The biggest mistake in venture isn’t losing money. It’s missing the outlier.”

He distinguished between two errors. The mistake of commission: you invest $10 million in a startup that zeros. That’s bad. The mistake of omission: you don’t invest in Google or Amazon. While the former costs a finite sum, the latter carries a near-infinite opportunity cost. “Venture is the most polarized possible economic field in which this is true,” he said. “You’re always much more worried about the mistake of omission than you’re worried about the mistake of commission.”

Andreessen is right. But he is unknowingly echoing a story that is two thousand years old, and whose deepest point is not about returns at all. It is about the opportunity cost applied to the self.

The Jurisprudence of Risk

The Parable of the Talents is usually read as a stewardship lesson. Deploy your gifts. Be productive. Don’t waste what God gave you.

This misses the legal tension in the story.

A master entrusts three servants with enormous sums: one talent was roughly twenty years of wages. He gives five to one, two to another, one to the last, each according to his ability. He departs. Two servants invest, trading and doubling their holdings. The third buries his in the ground.

When the master returns, he condemns the third servant, casting him into “outer darkness.” Why? The two who doubled are praised identically. The one who preserved perfectly is condemned.

The condemnation requires explanation. The third servant did not steal. He did not gamble. He did not spend. He returned exactly what he was given, and followed the letter of the law. He is cast into outer darkness.

Here is what most readers do not know: the Babylonian Talmud2 records that burial was the only legally safe way to guard money. A bailee who buried entrusted funds was exempt from liability for theft; robbery was far more common than ground damage, and burial was the mandated standard of care for entrusted property. The third servant was legally perfect and morally bankrupt.

He is condemned regardless.

The master does not say: you should have taken better risks. He says you ought to have put my money with the bankers — the ancient equivalent of: if this was your risk tolerance, you should not have been in the game at all. The problem is not the outcome. The sin is the refusal to engage. The mistake of omission.

But the deeper loss is not the master’s money. It is the servant’s atrophy. That is what Andreessen sees. What Jesus sees goes further.

The third servant did not just waste capital. He wasted the chance to become someone different. By burying the talent, he buried the pressure, the judgment, and the community that would have forged his character. Everything the encounter would have produced — the self-knowledge forged under pressure, the judgment developed through exposure to uncertainty, the character built through caring about something he could not fully control, the community formed in shared risk — went unrealized. He came back to the reckoning having refused every encounter that would have changed him. Unchallenged. Unchanged. Sterile. The talent was intact. He was not more. He was exactly what he had been.

And that, in a world designed for human beings to grow through engagement with an uncertain world, is the deepest possible failure.

The Internal Goods of Failure

For any venture investor reading this parable, one detail should unsettle them: if the parable were a spreadsheet, a 2x return would be a failure.

The two faithful servants each double their money. If the master simply wanted to double his capital, he would have bypassed the servants entirely and placed the funds in bonds. More striking: he gives identical praise to the servant who returned ten talents and the one who returned four. The percentage is the same. The absolute amounts are not.

The return is not what is being evaluated.

Alasdair MacIntyre’s After Virtue3, one of the most important works of moral philosophy of the 20th century, explains what is. MacIntyre distinguishes between two kinds of goods that any meaningful human activity produces. External goods are measurable, transferable, zero-sum: money, status, power. More for one person means less for another. Internal goods are entirely different. They are available only through genuine participation in a practice. They cannot be purchased, transferred, or obtained by watching from the outside. They belong only to those who have actually engaged.

MacIntyre is, in this sense, completing Aristotle. Eudaimonia4, flourishing, is not a state you achieve. It is an activity you inhabit. Something you do rather than something you have. The happiness is in the pursuit because the pursuit is where the internal goods live.

In venture capital, returns are external goods. They appear on spreadsheets, get distributed to LPs, convert into other things. The internal goods are what remain in the practitioner: self-knowledge forged under pressure; judgment refined through repeated exposure to uncertainty; character built by tethering your reputation to outcomes you cannot fully control; community formed in shared risk; and the pastoral depth — the capacity to sit with a founder at their lowest point and choose to stay — that only comes from having done it. These goods are accessible only to those who enter the practice fully.

The comedian Jimmy Carr5 arrived at this not from philosophy but from watching people — from his own journey out of poverty to success, from observing what difficulty produces in people versus what ease produces. His formulation is empirical: the external good, once possessed, is inert. The trust fund heir who inherits everything and finds himself mentally tortured; the person who chose the hard path and found it, despite everything, great. Having the thing is not where the life is. The life is in the becoming that occurs through the pursuit.

The two faithful servants are praised not for their percentage gains but because they entered the practice. They are capable of being trusted with more, not as a reward for productivity but because the practice developed their capacity to bear it. “Faithful over little, ruler over much.” Their growth was not a promotion. It was a transformation.

The servant who buried his talent committed the deeper mistake of omission: he refused to enter the practice. He refused to find out what he was capable of. He returned exactly what he was given, and consequently, he remained exactly what he was.

The Destiny That Chooses You

In Rounders (1998)6, Matt Damon plays Mike McDermott, a gifted poker player trying to bury his talent in law school. He sits in a bar with his professor, Petrovsky — a judge, a man of reckoning — after failing to show up when it counted. He expects to be admonished.

Instead, Petrovsky reveals his own history: he was supposed to be a rabbi. He followed his true self into the law anyway, despite the cost to his family, despite the social expectations. When McDermott asks whether he would make the same choice again, Petrovsky smiles.

“What choice?... We can’t run from who we are. Our destiny chooses us.”

This is the lived experience of the gift — not from the perspective where the assignment is clear, but from the inside, where it arrives as a pull rather than a command. McDermott keeps trying to bury his poker talent. It keeps finding him. He can delay the reckoning. He cannot escape what was given to him.

Notice what Petrovsky himself embodies. He was supposed to be a rabbi. He became a judge. But the wisdom he is deploying in this scene — in a bar, with a poker player afraid to follow his calling — is the wisdom he took from the Yeshiva. The talent didn’t stay buried. It followed him into a different container. The destiny chose him anyway. The gift refused to be fully interred.

The third servant is not just a bad investor. He is a man running from his own identity. He chooses the safe life over the great character. He refuses the destiny that chose him. And the reckoning at the end of the parable is not an arbitrary punishment delivered from outside. It is the moment of revelation: the moment when what the servant refused to become is made visible, when the opportunity cost of his personhood is finally accounted for.

The parable says he is cast into “outer darkness.” But the darkness is not the master’s imposition. It is the condition the servant himself created. To refuse formation is to hollow out the interior. The outer darkness is the outer shape of an inner darkness that was already there; the name of a life that refused its own becoming.

The master does not destroy him. He simply tells him what he already is: a person who had everything required to enter the practice and chose to remain a spectator.

What the Failures Taught Me

My time as an LP in Felicis and an investor at Horizons Ventures gave me the external goods: billion-dollar exits like Scopely and Shopify, and smaller but meaningful ones like Blue Vision Labs7 to Lyft. I learned what winning looks like from the inside. But the exits confirmed what I already thought I knew. What failure teaches you goes deeper than what success ever does.

The real internal goods were forged in the crucible of Proof of Capital, where as a founding GP we were first money into FTX8, which became one of the largest frauds in the history of venture capital; and years earlier at HTC, where from 2012 to 2015 I watched Apple and Samsung systematically dismantle our market share. Different kinds of failure. Both formative in ways no exit had been.

Bill Gurley9 at Benchmark has said for years that they never sweat the mistakes — only the missed opportunities. The commission mistake is bounded: you lose what you put in. The omission mistake is unbounded: you miss what could have been everything. The asymmetry between the two is not marginal. It is the entire structure of the venture business. The fund that fears commission more than omission has already lost.

The master in the parable is operating on this exact framework. He does not sweat the commission mistakes. The two servants who entered the market and could have lost everything — who took on the unbounded risk of trading in a first-century economy where caravans were robbed and debtors defaulted — are praised with no mention of return magnitude. The master explicitly invites commission. “Put my money with the bankers”, even passive deployment with interest counts. Anything except burial.

What the master sweats is omission. The buried talent is the omission mistake. And the master’s condemnation of the third servant is not disproportionate to the financial outcome. It is precisely proportionate to the asymmetry. The third servant chose the bounded zero of preservation over the unbounded possibility of deployment, and the master will not honor that choice.

Gurley and Jesus are running the same calculus: two thousand years apart, one in a Sand Hill Road conference room and one on the Mount of Olives. The convergence is not coincidence. It is the structure of how anything generative actually works, recovered independently across very different domains.

This is why I did not sweat the FTX loss the way the outside world expected me to. Not because it was small. Not because it did not hurt. But because the external good was bounded — a finite loss of capital, contained to that single bet — while never having engaged with the crypto paradigm at all would have cost me something categorically larger.

What I did not lose was the internal good. When an investment goes to zero or a startup fails, you do not lose everything. You lose only the external good — which was bounded to begin with. What you gain — what stays with you, what compounds across every subsequent attempt — is something no successful exit could have produced.

Three specific internal goods that I have only ever found in the failures.

First, failure teaches you human nature in a way success cannot. Why people do what they do. How decisions actually get made under pressure, rather than how they are reported afterward. How culture shapes execution far more than strategy does. How a founder responds when the company is dying tells you something about that person you could never have known when the numbers were going up. How a board behaves when the cap table is about to be wiped out reveals what each member actually believed all along. Success lets everyone perform the version of themselves they want to be. Failure strips that away. I have learned more about how human beings function — under stress, in groups, when the external goods evaporate — from the companies that went to zero than from any of the exits.

Second, failure gives you truth that nobody else has. When you commit capital, time, and reputation to an attempt and the market does not respond the way your thesis predicted, you have done something rare: you have generated data that nobody else has. Maybe the market was too early. Maybe the product was wrong. Maybe the founders were not the right founders. Whatever the answer, you now know something specific about that domain — about that exact intersection of timing, market, technology, and human capacity — that the people who watched from the sidelines do not. They have opinions. You have data. And the data changes you. You see the next opportunity in that space differently. You ask sharper questions. You spot the pattern earlier or you walk away faster. The truth you extracted from the failure is permanently yours. It is the kind of pattern recognition that cannot be acquired by reading or by observation. It is only acquired by deployment.

Third, failure reveals the worth of the attempt itself. The people who commit fully to a startup — who genuinely believe in what is being built — produce something worth celebrating even when the company fails. This is the part that is hardest to explain to people who have not been inside a real attempt. When a group of people believes in a product or a mission, when they take a risk together, when they show up every day and work toward something new without knowing whether it will work; there is something happening in that room that is independent of the outcome. It is its own form of communion. And when the attempt fails, that communion does not become retroactively meaningless. It was real when it was happening. The relationships that formed under that shared belief are real. The growth in each person was real. To have belonged, however briefly, to a group of people who fully committed to something together is one of the rarest privileges of a working life — and many people never experience it at all. The startup that fails has still produced this. And the people who lived inside it know how fortunate they were, regardless of what the cap table eventually said.

These are the three things failure teaches that success cannot: human nature in its uncovered form; truth extracted from data nobody else has; and the privilege of belonging to a group of people who fully committed to something together.

The mistake of commission, the failed investment, produces all of this. The mistake of omission, by contrast, produces nothing — not even the information that comes from a real loss. Just the perfectly preserved talent and a perfectly stagnant soul.

Sacks and the Soul Accounting

Rabbi Jonathan Sacks10 never directly commented on the Parable of the Talents; it is not his scripture. But his framework illuminates it more precisely than most standard readings.

Sacks argued that God deliberately left the world unfinished so that it could be completed by human work. A gift, therefore, is not given to be preserved. It is a mandate to enter the ongoing work of repair, tikkun olam11. Preservation is not stewardship. It is refusal.

The unequal distributions in the parable (five, two, one) are not divine favoritism. They are what Sacks called the dignity of difference12: the recognition that each person’s specific constitution represents a unique contribution, a theological law of comparative advantage. “By being what only I can be,” he wrote, “I give humanity what only I can give.” The third servant’s sin is the refusal of that specificity — the flattening of a particular gift into a generic act of safety.

This is Noah’s sin13. Noah walked with God — following, passive, waiting for instruction. Abraham walked before God — ahead, pioneering, entering uncertain territory first. The third servant is Noah. He practiced what Sacks called passive optimism: the belief that preserving the principal is enough, that waiting is faithfulness. But active hope, the conviction that together we can make things better, requires risk. Preservation requires none.

Sacks wrote: “Faith is not certainty — it is the courage to live with uncertainty.” The buried talent is a monument to passive optimism. The master’s return is its undoing.

The Talmud teaches that the first question asked in the world to come is not “how much did you make?” It is nasata v’natata be’emunah14 “Did you conduct your business faithfully?” Not: what were your returns? But: did you engage honestly with what was given to you? Did you enter the practice?

The master’s return is a soul accounting, not a performance review.

The Failed Founder Is Not the Third Servant

The third servant chose not to start. He had the talent in his hand, knew the game, and chose not to play because he was afraid. That is the precise sin the parable indicts.

A startup founder who loses everything is not the third servant. They entered the market. They exposed their capital — often their savings, their reputation, the years of their life — to the conditions of possible loss. By doing so, they already rejected the path of the buried talent. They built a team and a culture and an idea worth betting on. They found out things about themselves that are permanently theirs. The failure of the institution cannot take those goods back.

The startup founder who lost everything is not the third servant. The startup founder who never started is.

The mistake of omission is categorically worse than the mistake of commission, not just financially but developmentally. The person who never bets never finds out what they are. The person who bets and loses knows something new about themselves, has something to bring to the next bet, is more capable than they were before the first one.

If you tried and failed, you are further along than the person who never tried. The talent was not buried. It was deployed. The destiny chose you and you showed up. What came back may not look like a return. But the practice is already forming you, and the reckoning will account for it.

The Reckoning

The master is not waiting for a 2x return. He is waiting to see who you became while you were trying.

The redistribution at the end of the parable, “take the talent from him and give it to the one who has ten”, is not financial optimization. It is the recognition that capacity is built through practice. The one who participated, engaged, and developed the internal goods of faithful deployment can be trusted with more. Not because they generated the best return, but because they became the kind of person who can steward it.

This is the carried interest model of the kingdom. The return compounds in the person, not in the portfolio.

Petrovsky smiled when McDermott asked whether he would make the same choice again. What choice? The destiny had already chosen him. The only question was whether he would accept it — or spend his life burying and unburying the talent, delaying the reckoning, running from who he was.

I have been a founding GP who was first money into the largest fraud in venture history. I have sat on the board of billion-dollar exits. I have had more than ten exits and an uncountable number of failures. The external goods came and went. The internal goods — the self-knowledge, the covenant community of shared loss, the practice that lived on through the people, the pastoral relationships forged in failure — those are permanently mine. They are what the practice produced. They are what the reckoning will account for.

The question I carry is not what the returns were. It is whether I became, through all of it, the kind of person who can be trusted with more. Whether I took the journey. Whether I let it form me. Whether I accepted the destiny that kept choosing me.

The Demand

Andreessen: Fear the missed opportunity more than the failed attempt.

Jesus: Deploy the talent.

Carr: You can’t have an easy life and a great character.

Petrovsky: We can’t run from who we are. Our destiny chooses us.

What are you not starting? What is the conversation you are preparing instead of having? What talent are you keeping safe instead of putting into the market?

The mistake of omission is the one you will not be able to name at the reckoning, because it produced nothing — no data, no growth, no self-knowledge, no community forged in shared risk, no pastoral relationships formed in the crucible of failure. Just the perfectly preserved talent. The perfectly unchanged self. The opportunity cost of a personhood that refused its own becoming.

The destiny has already chosen you.

The only question is whether you will take the journey.

Subscribe now

This essay is part of the After Silicon series on faith, risk, and the formation of the soul. The second essay, The Antifragile Soul, argues that the structure of reality itself rewards those who enter the practice. The third, The Fusion of Horizons, explores what the encounters of the practice do to you — and why time changes everything you see.

The argument for structural separation now moves from historical precedent to architectural necessity across two critical domains: technology and money.

This piece builds on the concepts established in Part I and Part II: From Ploughshare to Spear:

1. The Separation of Technology and State

The innovation ecosystem is a spontaneous order — a Hayekian cosmos rather than a taxis. It is an emergent order that no single mind can design, fueled by entrepreneurial experimentation and open-source development that produce knowledge no central authority could ever possess. When the state attempts to direct this order, it inevitably substitutes the “pretence of knowledge1” for genuine discovery.

Waze serves as the primary example of technology pursuing its proper telos: solving the problem of navigation for a billion people because it improved their lives. However, Waze also serves as a warning. As a location data aggregator later acquired by Google, its utility was converted into a commercial intelligence resource. The dual-use character — where navigation data is indistinguishable from surveillance data — demands we distinguish between serving human ends and state ends more rigorously. We cannot assume commercial ownership guarantees this distinction. The tool remains the same, but the “standard of excellence” governing the work changes. When Waze served free users, it submitted to engineering’s internal standards. When the same technical substrate is redirected to “confidence-score” deportees, recommend targets, or build tax surveillance databases, it has been forced to serve a different master: state power.

Separating technology from the state does not mean an absence of governance. It means the state must perform its proper function: maintaining justice between spheres and preventing any actor — including itself — from absorbing the technological ecosystem into its own project of power. There is a compounding consequence to this fusion: when the Pentagon becomes the “prestige contract,” engineering talent follows. The consumer internet was built by curious, entrepreneurial people building tools for free human exchange. If the defining challenges of the next generation are restricted to weapons targeting and population surveillance, the talent that might have built the next communication platform will build the next kill-list algorithm instead. They are not “bad people”, but resources and incentives shape what gets built and who builds it. You cannot colonize a field’s best work and expect the field to remain what it was.

This corruption has a second, more dangerous dimension: what centralized control of AI does to knowledge itself.

Freidrich Hayek2’s argument about the price system was fundamentally about epistemology. The knowledge required for economic coordination is dispersed, local, and tacit; no central planner can aggregate it because that knowledge only exists through the act of independent exchange. Centralizing the planning function does not merely slow things down — it destroys the information that makes coordination possible.

The same argument applies to Artificial Intelligence.

When a single entity — whether a state or a monopoly platform — controls the model, the data, and the infrastructure, it centralizes the epistemic function of AI just as a planner centralizes prices. It does not merely gain power; it forecloses the diversity of interpretation and error-correction that makes any knowledge system trustworthy.

A single model, trained on a single dataset, and fine-tuned to a single organization’s values, is the AI equivalent of a single state-issued currency forced to serve all monetary functions at once: one monolith with no competition and no exit. It will produce coherent outputs, but they will carry the biases and the political noise of the controlling institution. By design, the system loses the diversity required to reveal its own distortions. This is not hypothetical; research shows that models trained predominantly on each other’s outputs suffer “model collapse” — a drift toward institutional legibility at the expense of everything peripheral or dissenting. It is Hayek’s calculation problem rendered in gradient descent3.

The remedy is structural. Open-source models on distributed infrastructure, trained on diverse data and evaluated by independent communities, are not an ideological preference; they are the precondition for AI that extends human understanding rather than amplifying whoever controls the weights. Just as a price system requires actors with “skin in the game”, a trustworthy AI ecosystem requires a genuine diversity of models and institutional homes. You cannot solve the “alignment problem” by perfecting the central planner. You solve it the same by preserving the conditions under which error can be identified and corrected.

The separation of technology from the state is a question of whether AI remains a tool for human knowledge or becomes the state’s ultimate instrument for making the world legible to power.

2. The Separation of Money and State.

Money is more than a tool for trade; it is a civilization’s primary mechanism for coordinating action across time and space. Prices encode dispersed information about scarcity and value that no central authority can replicate. As Hayek observed, essential knowledge never exists in a concentrated form, but only as “dispersed bits of incomplete and frequently contradictory knowledge ” held by individuals. When the state monopolizes and manipulates the money supply, inflation does not just raise prices — it distorts the entire structure of relative signals that guide human action. Without reliable economic calculation, Ludwig von Mises’s4 “guide through the ‘oppressive plenitude of economic potentialities,’” production becomes mere “gropings in the dark.”

The structural failure of the modern monetary system is not merely state control, but the lack of architectural separation between three distinct functions: the unit of account, the store of value, and the medium of exchange. Currently, a single politically-controlled instrument is forced to perform all three, and political incentives inevitably corrupt each role. Inflation targets distort the unit of account; quantitative easing destroys the store of value, and capital controls weaponize the medium of exchange. This is the same monolithic error found in centralized AI: forcing one institutionally-governed system to serve all epistemic roles with no competition or mechanism for error-correction.

The separation of money and state does not demand a single replacement instrument. It requires a principle: every monetary function must be insulated from political manipulation by architecture, not the shifting goodwill of central bankers. Bitcoin represents the first serious architectural proof of this principle — specifically for the store of value. Like digital gold with a fixed supply and no central issuer, it proves that an asset can hold value across time beyond the reach of political debasement.

Yet, as a single instrument, it remains an incomplete solution. In The Denationalisation of Money, Hayek did not argue for replacing one monolith with another; he argued for competing currencies. A healthy monetary ecosystem requires a plurality of instruments, each subject to the discipline of choice rather than political mandate. Bitcoin proved the principle of separation, but the rest — honest mediums of exchange and reliable units of account — are still being built.

Institutional history, from the gold standard to “independent” central banks, is a record of separations that were architecturally incomplete and therefore politically reversible. Separations not secured by an immutable structure will always erode. The direction is right. This leads to the third and most fundamental separation: energy from state. As the physical substrate for both technology and money, it is the most urgently needed architectural shift. It is the work Cold Electric is attempting to build — and the subject of an upcoming essay.

The Tree of Life

In The Technological Republic, Alex Karp5 argues that the fusion of technology and state is a prerequisite for civilizational survival. His premise is Hobbesian: power is the condition for all other goods, and states that refuse to weaponize technology will inevitably be dominated by those that do. This is a half-truth that adopts the logic of the very threat it claims to counter.

The real danger is the corruption of the practice itself. Waze was founded on the belief that technical expertise should solve human problems — that was sphere sovereignty in action, where technology submitted to its own internal standards. What Karp proposes is the inverse: consumer tech expertise crossing into the defense sphere and submitting to the state’s standards of lethality and surveillance. As Alasdair MacIntyre6 warned: when external goods like state power dominate, the internal goods of a practice are not merely neglected; they are corrupted. Engineers cease to be engineers and technology ceases to be a tool for human flourishing.

As Alexis de Tocqueville7 observed, the separation of church and state made religion more powerful by freeing it to be what it actually was. The solution to authoritarian tech-state fusion is not a “democratic” version of the same monolith. It is structural separation: technology freed to connect, money to encode value, and energy to power both. This is the only architecture capable of civilizational defense that does not become the monster it is defending against.

This refusal is not without cost; it entails a distinct competitive disadvantage. Yet the costs of such a refusal are bounded, whereas the costs of state-corporate fusion are structural and compounding.

James Madison8, the most rigorous theorist of institutional separation, spent decades analyzing the decay that follows state patronage. In his 1785 Memorial and Remonstrance Against Religious Assessments, his verdict on state-established Christianity was scathing: fifteen centuries of patronage had produced “pride and indolence in the Clergy” and “ignorance and servility in the laity.” Madison’s insight was that an institution accepting state support inevitably becomes a state instrument. This transformation is rarely a sudden betrayal; it is a gradual erosion. It occurs in stages — each concession small enough to rationalize, each dependency deepening the next — until the original mission of the institution is hollowed out.

The engineers currently accepting defense contracts face the same institutional gravity. They do not lose their integrity in a single moment of corruption. Instead they endure a slow replacement of the engineering telos with the narrow requirements of the contract: surveillance, lethality, and state legibility. When the forge is owned by the state, the smith no longer decides the shape of the metal.

I want to end with a piece of art, but the art is unintelligible without its context.

The Mozambican Civil War lasted fifteen years, from 1977 to 1992. It was a brutal proxy war of the Cold War: FRELIMO9 backed by the Soviet Union, RENAMO by Rhodesia and South Africa. The Mozambican people were the raw material upon which superpower ideologies were tested. One million died; five million were displaced. RENAMO did not treat starvation as a side effect of war; they used it as a weapon, systematically destroying farms and burning crops. Half a million children under five died of hunger rather than bullets. When the 1992 Rome Accords finally halted the slaughter, Mozambique was among the most devastated nations on earth. The war left behind only one thing in abundance: seven million guns.

These weapons remained buried in farms, hidden in floorboards, and slung across the backs of former combatants who had known nothing but conflict for their entire adult lives.

Bishop Dinis Sengulane, an Anglican leader who helped broker the Rome Accords, realized that peace could not take root in soil cluttered with hardware. He faced a population that could not plant food because they were still armed, still terrified, and still shaped by a culture where the gun was the instrument of survival. Sengulane did not reach for a political theory or economic frameworks; he reached for Isaiah 2:4: “They shall beat their swords into ploughshares, and their spears into pruning hooks.”

Sengulane read this text as policy rather than poetry. He envisioned a specific transaction: the technology of killing exchanged directly for the technology of growth. It was not a state-mandated confiscation or a melting down by an army. It was a voluntary, anonymous, one for one exchange. In October 1995, alongside the Christian Council of Mozambique, he launched Transformação de Armas em Enxadas (Transforming Arms into Tools).

The terms were absolute: an AK-47 for a plough; an RPG launcher for a hoe; a pistol for a sewing machine; a village’s cache of five hundred weapons for a tractor. There were no names recorded and no police present. “We are not disarming you,” Sengulane told the people. “We are transforming your guns into ploughshares so you can cultivate your land and get your daily bread.”

Sengulane understood that the prophecy was not about the absence of weapons, but about the reclamation of material. A gun is human craft redirected toward destruction; that same craft, returned to its proper telos—the sustaining of life—can feed a nation. Over nine years, this exchange decommissioned 600,000 weapons.

Four Mozambican artists — Kester, Hilario Nhatugueja, Fiel dos Santos, and Adelino Serafim Maté — eventually transformed these decommissioned tools into the Tree of Life. A three-meter-high sculpture made entirely of weapon components, the tree grows out of the very instruments of death it has superseded. Fiel dos Dantos, whose brother was kidnapped during the war, noted the irony: from these barrels and stocks, something was born that grows and develops. The sculpture, installed in the British Museum since 2005, is a literal fulfillment of prophecy.

This is not sentiment; it is architecture. It is the recognition that the technology of war is always capable of becoming something else, provided the institution wielding it is separated from the practice it corrupted. The steel did not change; its purpose did.

This logic now confronts our current horizon. The race to build Artificial General Intelligence is a contest between states — one constrained by energy, and the other by compute. Both are pouring capital and engineering talent into the most powerful instrument of persuasion, surveillance, and destruction ever conceived. AGI developed under these conditions will be a weapon by default. As James Madison cautioned regarding the corrupting intelligence of patronage, the institution that funds the craft ultimately dictates its practice.

We do not yet know the cost of this race, nor how many lives will be reshaped by tools built in the name of “civilizational survival”. But we must ask the question Isaiah asked of the AK-47: When this is over, what will the “ploughshare AGI” look like?

To beat this digital sword into something that serves human flourishing — the full realization of human potential freed from scarcity and state projects —requires a fundamental redirection. It means compute that extends understanding rather than surveilling it, and AI that serves the person rather than the institution.

This transformation requires “The Three Separations.” You cannot beat the sword into a ploughshare while the state still owns the forge. You cannot ask what AGI is for — in the deepest, most human sense — while the institutions building it have already answered that question with a defense contract. These separations are not the destination; they are the precondition. They represent the moment the weapons are laid down and the real work begins:

What shall we build now, and for whom?

Subscribe now

This concludes the third and last installment of the three-part series, From Ploughshare to Spear: On the Separation of State, Money, and Technology.

This piece builds on the concepts established in Part I of From Ploughshare to Spear:

The Reformation and What We Learned From It

In 1618, Protestant nobles in Prague threw two Catholic imperial governors out of a castle window. Both men survived — they landed in a dung heap — but the conflict they had come to embody did not end with them. What began as a religious dispute within the Holy Roman Empire metastasized over thirty years into a catastrophe that killed approximately eight million people, roughly a third of Central Europe’s population. Armies became self-sustaining machines of predation; entire cities were sacked and burned.

The core reason the Thirty Years’ War was so uniquely catastrophic was that there was no neutral ground. Religious identity and political identity were fused. Your prince’s confession was your confession. Your salvation was tied to your ruler’s theology. When the stakes of political conflict are infinite — when losing means not merely paying higher taxes but losing access to eternity itself — there is no basis for compromise. Every conflict becomes existential. Every peace is a pause between catastrophes.

The Peace of Westphalia in 1648 institutionalized the only available solution: it removed the question of which confession deserved imperial enforcement from the domain of questions that political power could or should resolve. This was not a clean modern separation of church and state; princes retained considerable authority over their subjects’ religion. However, it established the crucial precedent that religious uniformity was not a legitimate goal of imperial governance. The strategy was to take it off the table: govern what you can legitimately govern, and leave the rest to conscience.

The principle would take another century to work itself fully out, but the direction was set.

Alexis de Tocqueville, visiting America nearly two centuries later, observed the paradox that should anchor every essay written about institutional design: “They all attributed the peaceful dominion of religion in their country mainly to the separation of church and state”. The paradox: religion was more powerful in America precisely because it was separated from the state. In Europe, the fusion had corrupted both. The church became an instrument of political authority and lost its spiritual credibility, while the state acquired religious legitimacy claims it could not sustain. In America, freed from state entanglement, religion retained its genuine authority over conscience. The separation did not weaken the church; it freed the church to be what it actually was.

Roger Williams, who first proposed “a wall of separation” in 1644, made the argument with precision: The church becomes a tool of earthly power, and the state becomes an enforcer of theological uniformity. Neither can pursue its proper function. James Madison, surveying fifteen centuries of state-established Christianity, drew a grim balance sheet: “More or less in all places, pride and indolence in the Clergy, ignorance and servility in the laity, in both, superstition, bigotry and persecution.” The separation was not a concession to secularism; it was a recovery of what both institutions were for.

This is the model that should now be applied to money and to technology.

The Hobbes Problem and the Sphere Sovereignty Solution

Ben Thompson, writing in early 2026 about the standoff between major AI companies and the Pentagon over defense contracts, articulated the most serious version of the counter-argument: democratic accountability is sufficient. In this view, elected legislatures set law, companies operate within it, and the remedy for surveillance overreach is political, not structural. Thompson is not wrong that democratic accountability is necessary, but Tocqueville identified why it is insufficient: democratic majorities can colonize spheres just as effectively as monarchs can.

Everything Palantir does — from the IRS mega-database to the ImmigrationOS deportation tracking system — was authorized by elected officials following legal procedure. When Senators Wyden and Ocasio-Cortez wrote to Karp demanding answers about “a surveillance nightmare,” they were exercising democratic accountability. What they were not doing was addressing the structural condition that made the mega-database possible in the first place. Thompson trusts democratic procedure to constrain a tool it is simultaneously authorizing. The Holy Roman Empire made the same error.

Thomas Hobbes held that the sovereign must be absolute because divided authority produces only civil war. Every competing power structure threatens the Leviathan’s monopoly on legitimate force. Karp’s Technological Republic is Hobbesian in precisely this sense: technology must answer to the democratic state, and any claim to independence is a sovereignty no private actor has the right to assert.

The counter to Hobbes is not libertarianism, which merely relocates sovereignty in individuals. It is the tradition of sphere sovereignty, most fully articulated by Abraham Kuyper. Kuyper declared in 1880 that human life is not a uniform mass that the state organizes from above. It is “an infinitely structured organism” composed of distinct spheres — family, commerce, science, art, church — each with its own inherent authority, each answering to its own principles, and none deriving its legitimacy from the state. The state’s proper function is to maintain justice between spheres, not to colonize them. When the state exceeds this role, it becomes “an octopus, which stifles the whole of life.” As Kuyper put it: “Sphere sovereignty defending itself against state sovereignty — that is the course of world history.”

If sphere sovereignty is the institutional architecture, Alasdair MacIntyre’s moral philosophy supplies its inner logic. They converge on a single insight: practices have proper ends (telos), and institutions corrupt those ends when they substitute their own goods for the goods internal to the practice. Every genuine practice — engineering, medicine, science — has what MacIntyre calls internal goods: achievements accessible only by participating in the practice for its own sake. Engineering’s internal goods include elegant solutions and the satisfaction of solving real problems for real people.

Karp’s engineers would raise an objection: building effective systems for democratic governments is solving real problems for real people. Finding terrorists prevents casualties. But in this view, the internal/external goods distinction is lost. When engineers build deportation tracking systems and surveillance apparatus, they are not practicing engineering. They are practicing statecraft with engineering tools. The practice has been colonized, and its internal goods have been displaced by external ones: contracts, guaranteed revenue, and political access.

There is also a distinction between state and commercial surveillance that the current debate consistently elides: the question of exit. A user can delete their Google account; a citizen cannot opt out of the state. When commercial technology is absorbed into state administration, the disciplining mechanism of exit disappears entirely. The person scored by ImmigrationOS cannot choose a different government. This is not a difference of degree; it is a difference of kind.

MacIntyre’s warning is precise: “The ideals and the creativity of the practice are always vulnerable to the acquisitiveness of the institution.” Without justice, courage, and truthfulness, practices cannot resist the corrupting power of institutions. Intention cannot substitute for structure.

We have seen this corruption in real time through Karp himself. Early in Palantir’s history, Karp built a privacy architecture designed to limit data access. Those commitments eroded as contracts grew. A man who once said his biggest fear was fascism built the infrastructure that makes it technically feasible at scale. This is not hypocrisy; it is the corruption thesis working exactly as predicted.

Subscribe now

This concludes the second installment of the three-part series, From Ploughshare to Spear: On the Separation of State, Money, and Technology. The final entry, Part III, is published.

“They shall beat their swords into ploughshares, and their spears into pruning hooks. Nation shall not lift up sword against nation, neither shall they learn war anymore.”

— Isaiah 2:4

Silicon Valley, before the reversal.

I got interested in the internet because it connected people.

That sounds naive now, but it wasn’t naive in the 2000s, when the commercial internet was young and Silicon Valley was electric with the sense that something genuinely new was happening in the world. The animating belief was simple and serious: information wants to be free, networks want to grow, and when you connect enough people to enough knowledge, something transformatively good happens to human civilization. I believed it. Most of the people I worked alongside believed it. We were not wrong to believe it.

But I want to excavate the moment that belief took hold — because without a historical lens, we’re just looking at the wreckage without knowing what built it.

Silicon Valley did not emerge from a vacuum; it emerged from defense. The semiconductor industry was seeded by military contracts, shaped by Cold War priorities, and staffed by engineers whose first employers were often government-funded research programs. Fairchild Semiconductor to Intel, the entire substrate of the digital revolution grew in soil fertilized by the Department of Defense. The true origin story is not the garage; it is the Pentagon, and the engineers who took military funding and asked: what can this do for ordinary people?

Israel followed the same arc. At Horizons Ventures, I spent years backing founders who had come out of Unit 8200 — the intelligence corps that serves as the country’s entrepreneurial graduate school. These were people who spent their formative years building surveillance systems, breaking encryption, and processing intelligence at scale. When they left the army and asked: what can this do for ordinary people? They built cybersecurity companies. Analytics platforms. Enterprise software. Consumer applications.

Waze was one of the first Horizons Ventures’ investments in Israel.

At its core, Waze was location intelligence — the same technical substrate that enables precision targeting and drone strikes — applied to the question of how to get home faster. Its $1.1 billion acquisition by Google in 2013 was the catalytic moment for “Startup Nation”. It gave an entire generation the confidence that global consumer products could be built from Tel Aviv. Its technology was inseparable from the intelligence infrastructure that produced its founders.

This is the thread I have followed: defense to consumer, military to market, the spear becoming the ploughshare. The best minds of a generation took what states had funded for power and redirected it toward human connection. That was the dream that made the internet feel like liberation.

The Great Reversal

Something has reversed. The question being asked in the most serious rooms in technology — with massive capital and the deliberate recruitment of the best engineers — is no longer how to connect people. It is how to find them, track them, target them, and when necessary, destroy them.

The same location intelligence behind Waze now guides precision targeting. The same machine learning that recommends what to watch next now recommends who should be on a kill list. The data infrastructure that promised access to the world’s information is being assembled, at the request of governments, into a surveillance and weapons-targeting apparatus of historically unprecedented scale.

Defense funds are rising; the Pentagon is the growth contract. The most prestigious justification for this shift — articulated by Alex Karp, who trained under Habermas and wrote on the rhetoric of fascism — is that technology companies have a moral obligation to fuse with state power. Karp argues that the West’s survival depends on this, and that those of us who find it disquieting are indulging a luxury they cannot afford.

This is not the direction I got into technology to travel, nor, I suspect, was the intention of most who built the consumer internet. We went from connecting people to surveilling them; from defense to offense; from the ploughshare back to the spear.

The renaming of the Department of Defense to the Department of War in the current administration is more than a branding shift; it is a declaration of direction. We are witnessing the completion of a philosophical project that began when modernity replaced the question “what are we for?” with a sovereign Leviathan. For centuries, that state power was held in check by independent spheres it could never fully absorb: the church, the market, and the individual conscience.

But the contemporary state has achieved a fusion Hobbes never imagined. It has fused with money, which funds it without limit, and with technology, which allows it to see without limit. The result is not Hobbes’s Leviathan. It is something Hobbes never imagined: a Leviathan with omniscience, with infinite reach, with no institutional structure left standing between it and the individual.

This structure is apocalyptic not because it is inherently evil — it may be staffed entirely by people with good intentions — but because it recognizes no proper limits. Architecture, not intention, determines the outcome: a state that controls the money cannot be defunded; a state that controls the technology cannot be unseen. When the sovereign replaces conscience as the ground of order and fuses with the full stack of modern power, security is replaced by the engineering state’s fundamental tenet: treating people as aggregates to be managed rather than individuals to be addressed.

This is the crisis to which the Three Separations are a direct architectural response.

I have spent years asking what went wrong structurally. Why does the logic of tech-state fusion keep winning even when its consequences vindicate our concerns? And I have come to believe that the answer requires going further back than Silicon Valley, further back than Unit 8200, further back than the Cold War. We must answer the question every civilization eventually faces: What belongs to the state, and what belongs to the soul? Where does the state’s jurisdiction end and conscience, commerce, or community begin?

We have gotten this question catastrophically wrong before. We have also, once, gotten it right.

The argument I want to make is called the Three Separations, built on the lesson that current enthusiasts for “offensive tech” have failed to absorb.

The Problem Alex Karp Has Correctly Identified

Alex Karp represents the strongest version of the opposing view.

As CEO of Palantir and author of The Technological Republic, he is not a mere defense contractor retrofitting borrowed philosophy to his revenue stream. He is a man who spent years thinking about the relationship between ideas and power and reached a conclusion his peers in Silicon Valley are too comfortable to articulate.

His diagnosis: the West has confused its survival with its ends. Liberal values — pluralism, individual rights, freedom of conscience — are not self-sustaining. They endure because they are defended by hard power. As Karp argues, military capability, not moral superiority, explains the endurance of Western values.

His prescription: technology companies must re-engage with state power, not reluctantly but as a positive moral obligation. If democratic states refuse to develop surveillance and control tools, they will be dominated by states that do. The choice is not whether mass surveillance exists, but who controls it.

There is a deep irony here: Habermas, Karp’s mentor, spent his career diagnosing the “colonization of human life” by strategic action — systems that treat people as objects to be managed rather than persons to be addressed. Karp studied that diagnosis and then built the most powerful instrument of strategic action his generation has produced. This is the Hobbesian argument in its cleanest form. The threat Karp identifies is real, but his prescription — full tech-state fusion — is wrong in a way that requires a structural response. The error becomes visible only when you look at what it produces at scale, over time, in the society that has committed to it most fully.

The Engineering State and Its Logic

Dan Wang has lived inside that society for six years, through the Shanghai lockdown and Xi Jinping’s consolidation of power. His book, Breakneck: China’s Quest to Engineer the Future, is the best available response to Karp’s manifesto.

Wang’s central framing: “China is an engineering state, which brings a sledgehammer to problems… in contrast with America’s lawyerly society, which brings a gavel to block almost everything.” Engineers have quite literally ruled modern China; by 2002, all nine members of the Politburo Standing Committee were trained engineers. Xi Jinping studied chemical engineering; Hu Jintao built dams. By the metrics Karp cares about — the capacity to coordinate national resources and execute strategic objectives — the engineering state is formidable.

But Wang offers the sentence that should stop every reader cold: “The fundamental tenet of the engineering state is to look at people as aggregates, not individuals.”

This serves as the foundation upon which the argument rests. And what follows is a chronicle of what happens when a state fully absorbs technology into its project of power.

The one-child policy: Wang cites 321 million abortions and the sterilization of 108 million women. Enforcement was conducted, in his words, in ways where “wives and daughters were being sterilized in much the same way that farmers spayed their pigs.” In Shandong in 1991, a party secretary decreed zero births for a hundred days. Because 1991 was the year of the sheep, the episode acquired a second name: the slaughter of the lambs. The party secretary was promoted.

Wang’s structural observation is key: the one-child policy could only happen in an engineering state where there was no civil society to fight for legal protection against it. There was nothing in between the state and the individual.

Political scientist James C. Scott argued that the state’s persistent impulse is to make populations “legible” — to turn the particularity of persons into thin, manageable data. States have always wanted this. What Palantir sells is not a new temptation but a “legibility machine” that makes this aspiration achievable at a scale earlier states could only imagine. Karp has handed that capacity to states that already wanted it.

Then came zero-COVID. Drones flew around Shanghai broadcasting: “Repress your soul’s yearning for freedom.” This directive from a top mental health official was broadcast to 25 million people locked in their apartments. As Wang writes, the one-child policy allowed the Party to reach into women’s bodies; digital surveillance now allows it to control a person’s daily access to her shower. The neighborhood committees used for enforcement were never disbanded; they now call recently married women to ask about their menstrual cycles.

This is not a story about CCP ideology or specific historical contingencies particular to China’s development. Zero-COVID and the one-child policy were the outputs of a system in which no independent sphere retained the authority to say: this domain is not yours. Given the same tools and absence of countervailing institutions, any society will produce the same results.

Wang identifies the “Industrial Party” — Chinese bloggers who believe nation-states ruthlessly compete and the state must be organized around technology as the decisive weapon. He calls them “modern fascists.” They are, in essence, Karp’s thesis without the liberal-democratic guardrails. But Wang’s chronicle makes clear that guardrails are not self-maintaining. They require independent spheres — civil society, legal institutions — that tech-state fusion progressively erodes. Karp must explain not the initial democratic mandate, but the structure capable of resisting the mandate’s extension. Wang has shown us what happens when there is nothing there.

You cannot import the logic of the engineering state — viewing people as aggregates — and quarantine the consequences. It does not switch off for population control. Furthermore, surveillance infrastructure is asymmetric in time. It is easier to build than to dismantle. Despite years of scrutiny, the NSA’s capabilities are larger today than in 2013. Once a system exists, the arguments for preserving it are always more concrete than those for dismantling it. What looks like a reversible policy decision is, in practice, an architectural commitment. Karp’s guardrails face a ratchet, not a dial.

Subscribe now

This concludes the first installment of the three-part series, From Ploughshare to Spear: On the Separation of State, Money, and Technology. Subsequent entries, Part II and Part III, are published in order.

“Render to Caesar what is Caesar’s, and to God what is God’s.” — Gospel of Matthew 22:21

“I’m not a killer, but don’t push me.” — Tupac Shakur, Hail Mary

“Power tends to corrupt, and absolute power corrupts absolutely.” — Lord Acton

These three voices—street realism, political warning, and divine clarity—converge on a single question: what does moral power look like when it refuses to be naïve? The answer is the strength that sees the world as it is — riven by ambition, fear, and domination – understands its dangers without illusion, and nevertheless refuses to be ruled by them. This essay is about three figures who understood that. And one philosopher who didn’t.

I began paying closer attention to Tupac Shakur not merely as a musician, but as a thinker. When Tupac renamed himself Makaveli, it was not a branding decision. Artists change sounds all the time. They rarely change philosophies. Tupac’s choice to invoke Niccolò Machiavelli was declarative – a signal that he had arrived at a hard-earned understanding of how the world actually operates, and that moral language detached from that understanding offers no real protection to anyone.

So why Machiavelli? Why would a poet so attuned to injustice, vulnerability, and longing align himself with history’s most infamous analyst of power?

Machiavelli’s core insight was uncomfortable but true: power exists whether we acknowledge it or not. To ignore it is not righteousness. It is exposure. Tupac, navigating systems that punished naïveté and rewarded strategic clarity , grasped this instinctively. The name Makaveli was not a surrender to darkness. It was a refusal to be destroyed by it.

That refusal – to see the world clearly and still act morally within it – turns out to be the exact question the Gospels answer. And the answer is more Machiavellian than most readers expect.

The Trap

There is a moment in Jesus Christ’s public life when the atmosphere shifts from teaching to trial. The Pharisees approach with a question that sounds technical but is anything but: “Is it lawful to pay taxes to Caesar or not?” It is not a request for guidance. It is a political ambush, carefully staged and ruthlessly pragmatic.

The Pharisees intent was precise. Jesus had gathered too large a following to be ignored and too much moral authority to be tolerated. The tax question was designed to force an irreversible fracture.

Answer yes affirming the tax, and He alienates the crowds who see Him as a prophet standing against Roman domination.

Answer no rejecting it, and Rome has its charge – a revolutionary guilty of sedition.

Either way, He loses.

But the trap goes deeper than the binary. It also exposes the fault lines within the leadership itself. The Sadducees, bound to the Temple establishment, had already accommodated themselves to Roman rule. Their authority depended on cooperation within imperial power.

The Pharisees, by contract, drew influence from the people rather than from formal office. They resented Roman domination but feared the chaos of open revolt. Their challenge was therefore doubly motivated: to preserve their standing with the populace while signaling to Rome that they could manage dangerous ideas from within.

In asking about the tax, they attempted to weaponize the machinery of the empire without appearing to serve it. This was moral outsourcing, letting Rome do the silencing while the Pharisees preserved their own righteousness.

Jesus sees through this without hesitation. He asked for a coin. He looks at the image stamped upon it. He speaks: “Render to Caesar what is Caesar’s, and to God what is God’s”, one of the most extraordinary sentences ever spoken in public.

This is not evasion. It is authority exercised with precision. Caesar’s power is acknowledged, but radically contained. God’s sovereignty is affirmed without provocation. The empire is granted its due, but not the human soul. In one sentence, the Pharisees are denied the outcome they sought. Jesus asserts a higher sovereignty without ever naming Himself. The crowd understands. Rome has no grounds to act.

This is what Jesus meant when He instructed His disciples to be “wise as serpents and innocent as doves.” Wisdom without innocence becomes manipulation. Innocence without wisdom becomes naïveté. The Caesar reply is the most perfect embodiment of both held simultaneously.

Where Nietzsche Goes Wrong

Nietzsche dismissed Christianity as slave morality: a system born of resentment, sanctifying weakness because it lacked the strength to dominate. The noble values of ancient aristocracies – strength, pride, vitality – had been inverted by a slave class that could not seize power and so reframed powerlessness as virtue. Meekness, suffering, and self-abnegation were, for Nietzsche, not genuine foods but the disguised revenge of the weak upon the strong.

Consider what just happened in the Caesar scene. Jesus, faced with one of the most sophisticated political traps in recorded history – an ambush designed by experienced operators with full institutional backing – dismantled it completely. He named the machinery of the trap, reframed its terms, satisfied two competing audiences simultaneously, and walked away without yielding an inch of moral ground. Weakeness does not do that. Resentment does not do that. These are not moves of a slave psychology. They are the moves of someone operating from a position of deep interior authority.

Nietzsche’s error was categorical. He could not imagine power that does not need to prove itself through domination. His nobles demonstrate strength by conquest; his masters assert themselves through will and force. But Jesus reveals a different form of power entirely, one that is most visible in the moment it declines to be exercised.

These are the choices of someone who sees the knife clearly, could take it, and chooses not to. That is not slave morality. That is a higher order of mastery, one that Nietzsche (for all his genius) could not conceptualize because his framework was built on domination as the proof of strength.

What Machiavelli Taught & Where He Stopped

To see why this matters beyond theology, consider the enduring popularity of Robert Greene’s The 48 Laws of Power. Its appeal signals something real: a widespread fatigue with moral language that collapses under pressure, a hunger for honesty about how institutions, hierarchies, and human ambition actually function. Greene is Machiavelli translated into contemporary life. His readers are not cynics. They are people who have been burned enough times by naïveté that they are looking for a more durable map of reality.

Machiavelli’s insight – that ignoring power does not make it benign, it simply makes it unaccountable – is one of the most important intellectual contributions of the last five centuries. He was right about how the world works. Greene is right that those dynamics persist in every boardroom and institution today.

But Machiavelli stops where Jesus begins. Machiavelli teaches how power operates. Jesus shows how power should be ordered. The distinction is not minor, it is everything.

Jesus understands power at least as clearly as Machiavelli – the Caesar scene proves it – yet He refuses to grant it ultimacy. Power, in His hands, is always instrumental, never final. It is subordinated to covenant, dignity, and love. The person who reads Machiavelli and stops there has an accurate description of the world and no principle by which to judge it. That is the moral vacuum Greene’s readers are actually fleeing, even if they don’t have the words for it yet.

The Prince of Peace

This is why Jesus can be both the Prince of Peace and profoundly unsettling. He forgives, yet He confronts. He heals, yet He names hypocrisy with surgical clarity, calling the religious elite whitewashed tombs. He preaches mercy, yet overturns tables in the Temple. Jesus speaks in parables not to soften truth, but to divide those willing to hear from those invested in appearances. These are expressions of a single governing faculty: discernment. These are not the gestures of a harmless moralist. They are the actions of someone who understands that truth spoken carelessly can destroy, and truth spoken wisely can liberate. He knows when gentleness heals and when clarity must cut.

Peace, in this light, is not the absence of conflict but disciplined strength. Strength that knows when to confront, when to wait, and when restraint itself is the highest moral act.

The Prince of Peace is not naïve about power. He is sovereign over it.

And sovereignty of that kind is not passive. It is the most demanding form of authority there is.

Machiavelli taught us how the world works.

Tupac recognized that you must understand those rules or be destroyed by them.

Nietzsche urged us to seize power and prove our strength through it.

Robert Greene shows us how those dynamics persist in modern life.

But Jesus shows us something none of them could: how to inhabit a world shaped by power without being consumed by it. How to see every trap, understand every blade, and remain, through all of it, dangerously free.

That is not slave morality. That is the hardest thing in the world.

Subscribe now

“Software is eating the world.” — Marc Andreessen, 2011

“The internet is a copy machine. The digital economy is run on a river of copies. Unlike the mass-produced reproductions of the machine age, these copies are not just cheap, they are free.” — Kevin Kelly, 2008

“We wanted flying cars; instead, we got 140 characters.” — Peter Thiel

“Physics is the law. Everything else is a recommendation.” — Elon Musk

Every technological revolution produces the same illusion: that the new layer has permanently displaced the old one. Steam made muscle obsolete — until factories needed more of it to feed the machines. Electricity made gas lighting irrelevant, until electrification created demand for more energy than gas ever consumed. The internet made physical retail seem doomed, until e-commerce created more warehouses, more logistics workers, and more physical infrastructure than traditional retail ever required.

In 1865, the economist William Stanley Jevons observed that more efficient steam engines did not reduce coal consumption — they increased it, because efficiency made steam power economical across a far wider range of applications. What appears to be substitution is amplification. This is the Jevons Paradox: efficiency in the use of a resource tends to increase, not decrease, total consumption of that resource.

Software is experiencing its own Jevons moment. Understanding it correctly is the difference between seeing what comes next and being surprised by it.

When Bits Ruled

The last era belonged to bits.

Software was powerful because it escaped physics. Code could be copied perfectly, transmitted instantly, and scaled globally at near-zero marginal cost. The bottleneck of the industrial economy — the cost of replicating and distributing things — simply collapsed. Music, maps, photography, retail, finance, media: anything that could be turned into bits was eventually devoured by software. Intelligence, once embodied in people and organizations, was encoded into algorithms and deployed everywhere at once.

This was not merely a business story. It was a shift in where scarcity lived. For most of human history, scarcity lived in physical production — in land, labor, materials, and manufacturing. The software era moved scarcity upward, into minds: into the engineers who wrote the code, the designers who shaped the interfaces, the entrepreneurs who saw the abstractions before others did.

Then artificial intelligence arrived. And the scarcity moved again – this time, downward.

The Jevons Paradox of Intelligence

AI does to software what efficient steam engines did to coal. It does not reduce the consumption of software — it expands it, by making software cheap enough to apply everywhere that was previously uneconomical. One engineer today, augmented by AI, can build what previously required a team. Workflows that could never justify custom software now can; applications that were too expensive to maintain no longer are. The leverage of writing software has increased so dramatically that more software will be written in the next decade than in all previous decades combined — not because the prediction is provable, but because the economics of production have never pointed more sharply in one direction.

This is Jevons operating at the level of intelligence itself. But the Jevons framing, followed to its conclusion, reveals something uncomfortable: what expands is consumption, not the value at the production layer. Coal consumption expanded after the efficient steam engine — but the value didn’t accrue to coal. It accrued to the machines, the factories, the railways, and the industrial systems that consumed coal as an input. Coal became infrastructure — essential, abundant, and impossible to price at a premium.

Software is becoming coal.

Essential, yes. The entire AI revolution runs on it. But its value, over time, is concentrating not in the resource itself but in the physical architectures that transform energy into intelligence.

When intelligence becomes abundant, the value migrates to whoever controls the substrate that intelligence runs on. And that substrate is governed entirely by physics.

AI Is Not Digital. It Is Industrial.

Artificial intelligence feels intangible. We encounter it through screens, chat interfaces, and APIs. But intelligence at scale is not abstract. It is industrial.

Every unit of AI must be computed on silicon, stored in memory, cooled through air or liquid, and powered by electricity. Models do not drift through the cloud. They sit inside machines, inside buildings, connected to grids. Training a large model consumes more electricity than many mid-sized cities. Running inference at global scale demands a reliability and density of power delivery that most national grids were designed around fundamentally different load profiles to provide.

AI does not escape physics. It collides with it.

The constraint on intelligence is not algorithmic. Algorithms improve quickly. The constraint is energy density, thermal management, silicon geometry, and the reliability of power delivery at scale. These are engineering problems that take years and capital to solve — and they cannot be iterated away behind a keyboard.

You cannot prompt a semiconductor fab into existence. You cannot auto-generate a battery factory. You cannot deploy a model where the grid cannot follow. The more capable AI becomes, the more it runs into the hard limits of the physical world. This is why the returns to physics are accelerating, not decelerating, as AI improves.

Hard Hats Over Soft Heads

The Economist recently noted a quiet but unmistakable shift in where value is concentrating:

“Everywhere hard hats are triumphing over soft heads. Asset-heavy firms are outperforming ones which investors now fear have no valuable assets at all, like software companies.”

The evidence they pointed to was General Electric — a company whose rise, fall, and resurrection is one of the most instructive industrial stories of the past century.

GE’s fall was, in essence, a bet against physics. Under Jack Welch, the company that had built its identity around manufacturing — jet engines, gas turbines, power systems, industrial machinery — spent two decades migrating toward financial abstraction. GE Capital, its financial arm, contributed nearly half of the company’s earnings by 2005. The logic was compelling on paper: financial engineering was scaled faster and yielded higher margins than physical engineering. Bits and balance sheets over atoms and turbines.

The 2008 financial crisis nearly destroyed the company. GE Capital, built on short-term funding markets and complex asset structures, almost sank the entire enterprise. The physical businesses — the ones that made real things under real constraints — survived.

What followed was not a return to the past but a clarification of where durable value had always lived. GE broke into three companies. GE Aerospace and GE Vernova, the power and industrial spinoffs, now dominate industries that the AI era has made newly strategic. GE Aerospace exerts what The Economist called herculean pricing power on jet engine spare parts. That moat is built not from code but from installed base, regulatory certification, metallurgical expertise — and the irreplaceable cost of grounding a plane. GE Vernova is experiencing explosive demand for gas turbines and electrical infrastructure as technology companies build the data centers that AI requires.

Their combined market value, as of early 2026, is now five times what GE was worth before the breakup. The physical businesses, once considered the legacy drag on a financial conglomerate, turned out to be the durable core.

The lesson is not about the failure of finance or abstraction. It is that physical systems, governed by physical constraints, accumulate moats that are qualitatively different from software moats — and that those moats become most visible precisely when the informational layer above them flattens and commoditizes.

The Stack Has Inverted

The old mental model placed software at the top of the value stack, drifting above the physical world. The physical layer was legacy — something to be disrupted, disintermediated, or ignored. Value flowed upward toward abstraction, toward networks, toward code.

That model is inverting.

Energy forms the foundation; hardware sits above it; compute follows; intelligence emerges only once those layers are reliably in place. Software remains essential — but it is a layer in a stack, not the stack itself.

Software rides on physics. Physics does not ride on software.

This inversion is already visible in the strategic behavior of the largest technology companies. Google, whose dominance was built on search algorithms, recognized years ago that AI would pull value down the stack. Its investment in the TPU — custom silicon developed internally — was an acknowledgment that at scale, intelligence is constrained less by algorithms than by the efficiency of the physical substrate. Amazon Web Services is, at its core, an energy and hardware business — one of the world’s largest builders of data centers, custom chips, and power infrastructure — that happens to present itself through a software interface. Nvidia’s value is not just its code. It is the silicon geometry, thermal architecture, and manufacturing relationships that take years to replicate.

The pattern is consistent: wherever intelligence becomes operationally decisive, advantage collapses into atoms and watts.

Atoms Return. Watts Become the Chokepoint.

Atoms are unforgiving in ways that bits never were.

They demand factories, materials science, supply chains, tolerances, safety systems, and time. They do not scale overnight, and they do not forgive shortcuts. They reward patience, capital discipline, and operational excellence accumulated over years. A company with a superior algorithm can be copied in months. A company with a superior manufacturing process, a certified supply chain, and a proven safety record takes a decade to replicate.

If atoms are the body of the new infrastructure, watts are its bloodstream.

The twentieth century was organized, economically and geopolitically, around oil. The twenty-first century is being organized around kilowatt-hours: who can generate them reliably, store them safely, deliver them at the density and uptime that intelligence at scale requires. Bitcoin’s proof-of-work protocol makes this conversion literal: it is electricity transformed directly into monetary value, with no abstraction between the watt and the asset. The most abstract-seeming currency in the world turns out to be one of the most energy-hungry things humans have ever built.

Grids strain under new loads. Interconnection queues stretch years into the future. Heat accumulates in data centers faster than conventional cooling can remove it. The energy constraint on AI is not a temporary bottleneck to be engineered away — it is a permanent feature of what intelligence at scale actually costs.

This is not a problem to be solved. It is infrastructure to be built.

Taiwan: The Switzerland of Technology

If there is one place on earth where the return of atoms is most legible, it is Taiwan.

For decades, Taiwan built its identity around something the software era consistently undervalued: the disciplined, high-tolerance manufacture of physical systems. Its OEM and ODM ecosystem — the dense network of contract manufacturers, component specialists, precision engineers, and supply chain orchestrators — made it the invisible backbone of the global technology industry. Every iPhone, every server, every semiconductor node that powered the software era passed, at some point, through Taiwan’s industrial capacity. The island did not write the software. It built the machines that ran it.

That positioning, once dismissed as a commodity role in an era that celebrated abstraction, has become one of the most structurally durable advantages in the global economy.

Taiwan is the Switzerland of technology — not merely neutral ground, but trusted in the one sense that matters for the physical technology stack: trusted ground. Switzerland’s historical value was not its military power but its institutional reliability: a place where adversarial parties could store value, settle transactions, and build things that required trust across borders. Taiwan holds the same position in global manufacturing. Its relationships span geopolitical divides. Its quality standards are non-negotiable. Its engineers have spent generations accumulating tacit knowledge — the tolerances, the process discipline, the supply chain intuition — that no competitor can replicate quickly, regardless of capital.

This trust premium is becoming more valuable, not less, as hardware moves back to the center of strategic competition. Advanced semiconductors, battery energy storage, AI data center infrastructure, autonomous systems, drones — the things the world needs most urgently are precisely the things that require manufacturing credibility built over decades. Taiwan has it. Most of the world does not.

The semiconductor story is the most visible layer. TSMC is the irreplaceable node in the global chip supply chain, and virtually every major AI model trained or deployed anywhere in the world runs on silicon that flows through Taiwan. But the infrastructure story runs deeper still. The physical data centers being built for the AI build-out are not being assembled by software companies — they are being built by Taiwanese OEM and ODM manufacturers who have spent decades optimizing yield, minimizing failure rates, and building at scale under extreme quality constraints. Quanta, Foxconn, and Wistron are constructing the servers, racks, and infrastructure systems that the world’s AI capacity actually runs on. The same manufacturing culture that put a smartphone in every pocket is now building the physical substrate of global intelligence.

And the frontier is expanding further than data centers. Autonomous systems, drones — the hardware that defense planners increasingly consider decisive in contested environments — demand exactly the kind of trusted, high-volume, precision manufacturing that Taiwan has spent generations developing. This is where the Switzerland analogy sharpens into something more consequential: the ability to manufacture trusted hardware at scale is no longer merely a commercial advantage. It is a geopolitical one.

Taiwan does not need to choose between chips, batteries, and drones. It has already built the industrial culture that all three require. The return of atoms to the center of the global economy is, for Taiwan, less a disruption than a vindication.

From Atoms to Bits to Watts

I have lived this transition from the inside.

In 2011, I was at HTC at the peak of its powers — the largest consumer brand to emerge from Taiwan. What I didn’t understand then was that the small lithium-ion battery we sourced from ATL — a supplier whose alumni would go on to found CATL, now the dominant battery manufacturer in the world — was not merely a component in a smartphone. It was a preview of the most strategically consequential object in the AI infrastructure build-out today.

I spent a decade after that investing in software, in the fullness of the bits era. Then AI arrived and began to do to software what software had done to everything else.

Now, at Cold Electric, I am building LFP battery energy storage systems and containerized data centers. The physical infrastructure that intelligence actually runs on. Containerized energy and compute systems treat power the way modern software treats services: modular, deployable, and resilient – infrastructure that moves to where intelligence is needed, rather than waiting for a grid designed for a different century.

The atoms-to-bits-to-watts arc is a map of where value has moved — and is moving — across the largest technological transition of our lifetimes.

The End of the Software Illusion

Software ate the world when intelligence was scarce.

AI makes intelligence abundant. Jevons tells us what follows: abundance at one layer amplifies consumption and migrates value to the constraining layer below.

The constraining layer below software is physics.

Physics does not iterate. It does not move fast and break things. It simply governs — with the same indifference it has always shown to every era that abstracted itself away from material reality.

This thesis is no longer contrarian. The Economist is writing it. Equity markets are pricing it. The labor data is already confirming it — the occupational categories most protected from AI displacement are, with striking consistency, the physical ones: electricians, equipment operators, construction managers, manufacturing technicians. The knowledge work roles the software era made prestigious are the most exposed. The transition is not only economic. It is redistributive in ways that cut against two decades of received wisdom about where skilled work lives.

The window for acting on it ahead of consensus is not closed — but it is narrowing. The entrepreneurs and investors who recognized the software era early built the defining companies of the last twenty years. The ones who recognize the physics era early will build the defining companies of the next twenty.

The Economist is right: hard hats are triumphing over soft heads. But this is not a cultural shift or a market rotation. It is physics reasserting the priority it never actually surrendered.

Software had its moment.

Hardware is eating the world.

Part I: WalkmanからMegawattまで

テンポ、スケール、そして拡張の規律

「直感的思考の特徴は、それが迅速かつ労力を要さず、自動的に生じることにある。」 — Daniel Kahneman

「Softwareが世界を飲み込んでいる。」 — Marc Andreessen

「熱狂の後にこそ、相乗効果（シナジー）が訪れる。」— Carlota Perez

文明は、互いに矛盾する「時計（クロック）」の上に築かれている 。私はそのうちの一つを Silicon Valley で学んだ 。そこにおいてスピードとは単なる性格特性ではなく、構造的な要件であった 。ネットワーク化された市場では、躊躇による不利の蓄積は、誤りによるリスクの増幅よりも速い 。製品が7割完成していれば出荷し、不具合があればパッチを当てる 。Reid Hoffman の提唱した「ブリッツスケーリング（電撃的な規模拡大）」は単なる流行語ではなく、一種の生態学的適応であった 。

流通コストがゼロに等しくネットワーク効果が支配する世界では、最初に領土を占拠した者が勝者となる 。市場シェア、アテンション、そして資本はすべて同じ複利曲線を描く 。Marc Andreessen はこれをさらに鋭く抽出し、「Software が世界を飲み込む」と表現した 。流通がコードへと集約され限界費用がゼロに近づくとき、勝者は往々にして「最初かつ最速でスケールした企業」となるのである 。

意思決定の哲学さえも、速度（ベロシティ）を中心に再設計された 。Jeff Bezos は、取り消し可能な決定と不可能な決定を区別し、ミスを挽回できる場合には迅速に動くようチームを促した 。100％の情報を待って機会を逸するよりも、70％の情報で行動する方が望ましい場合が多かったのである 。この環境において、失敗は致命傷にはならない 。コードは刹那的であり、サーバーは再配分される 。バグの影響が存続に関わることが稀であるため、システムは加速を奨励するのである 。

テンポはスピードではない

数年後、プロバスケットボールチームの運営に携わる中で、私はより微細な真実を理解し始めた 。スピードとテンポは同義ではないということだ 。若いチームはしばしば「動き（モーション）」を「制御（コントロール）」と勘違いし、自滅するまで全力疾走してターンオーバーと疲労を繰り返す 。それはリズムを欠いた加速に過ぎない 。

一方で、Luka Dončić のような選手がいる 。彼はコート上で最速ではないが、試合を支配している 。他者が急ぐところで静止し、ディフェンダーが弛緩した瞬間に加速する 。彼は時計と競うのではなく、時計を操るのだ 。これが「テンポを支配する」ということである 。

テンポとは「いつ速く動くべきか」を決定することである 。イノベーションも同様だ 。Silicon Valley の時計は速度に報いる 。そのアリーナが求めたのは、先行者利益とネットワーク効果であったからだ 。しかし、すべてのアリーナが同じ時計で動いているわけではない 。誠実さや整合性（インテグリティ）こそが報われるアリーナも存在するのだ 。

消費者の時代：オブジェクトからプラットフォームへ

この変化は、AI やメガワット規模のデータセンターが話題をさらうずっと前から現れていた 。我々はその転換を、Walkman から iPod への移行に見ることができる 。

Sony は Walkman を「工芸品」として作り上げた 。それはハードウェアの粋を極め、ポケットに収まるサイズに凝縮された傑作であった 。それは音楽だけでなく、個人のアイデンティティを運ぶものであった 。しかし iPod の登場とともに、Apple は価値が「モノ（オブジェクト）」から「プラットフォーム」へと移行したことを見抜いた 。Steve Jobs の下で Apple は、デバイス、ソフトウェア、流通、ストアフロントを統合した「システム」を構築したのである 。iTunes はライブラリを同期し、後の App Store は産業全体を同期させた 。それらは単なるソフトウェアではなく、「エコシステム」であった 。

価値はオブジェクトからネットワークへと移行した 。これはエンジニアリングの敗北ではなく、経済構造の変革である 。流通が瞬時になり限界費用がゼロに近づくとき、重心は移動し、スケールそのものが戦略となる 。これを理解した企業がプラットフォームを築き、ハードウェアのみに固執した企業は苦戦を強いられた 。ここではスピードが決定的要因となったのである 。

バッテリーと産業拡張の論理

同様の物語はエネルギー貯蔵の分野でも繰り返された 。プラットフォームの論理が、最終的にハードウェアへと浸透したのである 。

Panasonic は、品質管理やプロセスの厳格さ、段階的な改良といった日本の産業規律の伝統を継承してきた 。その電池は長年にわたりデバイスや電気自動車（EV）を支えてきた 。そのエンジニアリングは慎重かつ精密、そして組織的であった 。

しかし、EV革命は新たな競争の地平を切り拓いた 。化学組成と同じくらい、製造規模、サプライチェーンの統合、資本へのアクセスが重要になり始めたのである 。CATL が猛烈な勢いで拡張し、驚異的な速度でギガファクトリーを建設し始めたとき、彼らは単に化学を改良しただけではなかった 。彼らは生産能力を「ブリッツスケール」したのである 。契約は倍増し、市場シェアは複利的に拡大した 。

彼らは、グローバルなエネルギー転換においては「規模が保護（プロテクション）」となり、「拡張が絶縁（インシュレーション）」となることを理解していた 。ブリッツスケーリングの論理が Software からハードウェアへと移行し、時計の針は加速したのである 。

環境への適合としての「速さと遅さ」

「速いこと」と「遅いこと」を対極として捉えがちだが、そうではない 。それらは環境への適合（Environmental Fit）の表現である 。

「速い」イノベーションは拡張を最適化する 。それは、戦略を撤回でき、バグをパッチできるという「可逆性」を前提としている 。変動性を許容することで、リーチを最大化するのだ 。対照的に「遅い」イノベーションは「整合（アライメント）」を最適化する 。相互依存性を前提とし、展開（デプロイメント）前にリスクを徹底的にフィルタリングする 。それは統合（インテグレーション）を最大化するためである 。

この区別は構造的なものであり、哲学的なものではない 。「失敗がアップデート」で済む領域では速いイノベーションが栄え、「失敗が大惨事（カタストロフィ）」となる領域では遅いイノベーションが繁栄する 。コンシューマーアプリでのミスは些細な不便に過ぎないが、サプライチェーンでのミスはリコールを招き、エネルギーシステムでのミスは「停電（アウトエイジ）」を引き起こす 。テンポは、その結果（コンシークエンス）の重みに合致しなければならない 。

スピードの隠れた前提

Silicon Valley の速度に対する自信は、一つの暗黙の前提に基づいている。それは「ほとんどのミスは致命傷にならない」というものだ 。機能が失敗すればロールバックし、市場が縮小すれば資本を再配分する 。電力、帯域幅、ロジスティクスといった「基盤」が安定しているからこそ、システムはこれらのショックを吸収できるのである 。

しかし、イノベーションがインターフェースの裏側へと移動するにつれ、エラーの許容範囲（マージン）は狭まっていく 。テクノロジーが病院、輸送ネットワーク、工場、そして都市全体の電力を消費する AI クラスタを動かすようになると、アライメントを欠いたスピードはもはや優位性ではなく、脆弱性（フラジリティ）へと変貌する 。

熱狂から展開（Deployment）へ

Carlota Perez は技術革命を二つの段階に分けて説明している 。第一段階は「導入（Installation）」であり、資本が制度を追い越す熱狂、投機、急速な構築の時代である 。過去20年間はこの段階に定義されていた 。Software は驚異的な速度でスケールし、インターフェースが既存産業を飲み込んでいった 。

しかし、我々は今「展開（Deployment）」の段階に入りつつある 。テクノロジーが社会の中核を再構築するフェーズである 。AI はもはや単なるデジタル層ではなく、産業的な電力消費者となった 。データセンターは単なるサーバー群ではなく、巨大な産業施設（ファシリティ）である 。議論の焦点はユーザー成長から、メガワット、冷却システム、サプライチェーン、そして電力系統（グリッド）の安定性へと移行した 。テクノロジーが物理世界を形作るとき、熱狂は「統合（インテグレーション）」へと譲歩しなければならない 。

スクリーンの下の時計

コンシューマー向けインターネットの成功は、我々に「すべてが Software のスピードで動くべきだ」という錯覚を与えた 。デジタル領域においては正しいその信念も、インターフェースの下にある物理的制約（フィジカル・コンストレインツ）を無視することはできない 。

電力網。製造プラント。原材料の輸送路 。これらのシステムは異なる時計（クロック）で稼働している 。彼らはイノベーションを拒んでいるわけではないが、変動（ボラティリティ）を吸収することはできないのだ 。そのコストは数十年単位で測定され、その失敗は数世代にわたって記憶される 。

過去20年間は速度とスケールが報われた時代であった 。しかし次のフェーズで求められるのは、より困難な「キャリブレーション（調整）」である 。アプリはエラーを許容するが、電力系統（グリッド）はそうではない 。

イノベーションがスクリーンからメガワットへと移行するとき、テンポはもはや美的選択ではない 。それは構造的な「責務（マンデート）」となるのである 。

Part II: Nintendo Energy

主権、エコシステム、そして「中間層」の在り方

「遅れたゲームは最終的には良いものになるが、急いでゲームをすると永遠に悪いものになる。」 — 宮本 茂

急がば回れ (If you are in a hurry, take the long way around.)

モバイルゲームが急成長を遂げていた頃、私は Silicon Valley のあるベンチャーキャピタルのオフィスにいた。そこでは、「スマートフォンがゲーム業界のすべてを飲み込む」というシンプルかつ自信に満ちたテーゼ（命題）のもと、新たなファンドが形成されつつあった 。

その論理は明快だった。全人類のポケットにハードウェア（端末）があり、流通の摩擦はなく、収益化モデルは再定義される 。そんな時代に、一体誰が再び「コンソール（専用ゲーム機）」など作るだろうか？

その視点から見れば、Nintendo が Mario や Donkey Kong を iPhone へ移行させることを拒む姿は、頑迷で、自滅的ですらあるように映った 。当時の株価には、「波はすでに過ぎ去った」と信じる投資家たちの不安が色濃く反映されていた 。

その主張は、Silicon Valley が「スケール」に関して信奉するあらゆる真理と一致していた。「プラットフォームが勝利する」「流通こそが宿命である」 。インターフェースを制する者が市場を制する。数十億のユーザーを抱える新たなチャネルが開かれたなら、そこへ参入するか、さもなくば取り残されるかの二択である、と 。

しかし、Nintendo は一線を守り抜いた。自らの世界を、他社の流通レイヤーへと明け渡すことはしなかったのである 。業界全体がピボット（転換）する中で静止し続けることは、稀有な規律を要する 。

私は今でも、彼らがどうやってその一線を守り切ったのか完全には理解できていない 。外部からの圧力は計り知れないものだったはずだ。iPhone は世界最強の流通プラットフォームとなり 、アナリストは包囲網を敷き、投資家は苛立ちを募らせた。四半期ごとの指標は、一様に「移行せよ」と囁き続けていたのである 。

市場は「忍耐」を罰し、株主が「自制」を称賛することは稀である。しかし、多くの場合、自制こそが「コヒーレンス（整合性）」を保つ唯一の道なのだ 。

ついに Switch が登場したとき、それは市場への反応（リアクション）ではなく、一つの「継続性」として現れた 。そこには、数十年にわたり Nintendo を定義してきた本能が息づいていた。遊びをスペック（機能群）ではなく「体験」として捉え 、喜びを仕様書ではなく「戦略」として捉える本能である。Nintendo は拡大よりも「保全」を選んだ。それはノスタルジーではなく、内なるリズムを守るための戦略であった 。刹那的なダウンロードが溢れる世界で、彼らは「完結した世界」を構築することを選んだのである 。

巨大な波が通り過ぎるのを眺めながら、そこへ飛び込まないでいるためには、並外れた規律が必要となる 。

岐路：京都 vs. 台北

その数年前、私はこの緊張感の「高速クロック（Fast Clock）」版を目の当たりにした 。HTC にいた頃、我々はコンソールとスマートフォンの溝を埋めるプロトタイプを製作した。デュアルスクリーン、物理コントローラー、そして遊びに重ね合わされた接続性 。端的に言えば、HTC による Nintendo DS フォンである 。

私はそのプロトタイプを手に、涼やかな空気が流れ、質素ながらも精緻な建築が並ぶ Kyoto の路地を歩いたのを覚えている 。落ち着いた雰囲気の会議室で、そのアイデアは熱心な好奇心と、ハードウェアへの深い理解に対する敬意をもって迎えられた 。日本の取締役会で熱意が声高に語られることは稀であり、彼らはまず「吸収」する。ある取締役が身を乗り出した。そこには確かな好奇心があった 。それは再度の会合を設けるに値するものだった 。

一瞬、二つのエコシステム——Taiwan のハードウェア技術と日本の創造的な世界——が交差する予感があった 。

しかし、イノベーションが一つの部屋だけで決まることは滅多にない 。Taipei に戻り、プロトタイプが別の会議室に持ち込まれると、下された審判は迅速だった 。当時の CEO はデュアルスクリーンのユーザビリティを認めず、その製品が出荷されることはなかった 。これは、私が先駆的に取り組んだアイデアの中でも、もっとも心残りのあるものの一つだ。それが成功したはずだからではなく、決して選ばれることのなかった「分岐点」を象徴していたからである 。

今振り返れば、Nintendo が iPhone への移行を拒んだのは、頑迷さではなく「明晰さ」ゆえだったことがわかる 。自らの世界を他者の Substrate（基底層）の上に接ぎ木してしまえば、自らのテンポを決定する力を失うということを、彼らは理解していたのだ 。一度選んだ「主権」というものは、中途半端に維持できるものではない 。

空洞化する中間層 vs. 濃密なエコシステム

音楽業界もまた、同じ緊張感の静かな側面を露わにした 。ストリーミングが Spotify と同義になる以前から、アジアではすでにサブスクリプションモデルの実験が始まっていた 。私は初期の合法ストリーミングプラットフォームである KKBox に投資した。モデルは機能した。ユーザーは月額料金を支払い、海賊版は減少し、カタログは拡大した 。

ストリーミングの勝利は必然に思えた 。しかし、スケールには重力が伴う 。グローバルなストリーミングモデルは「リーチ」を最適化した。関心を単一のインターフェースに集約することで、アクセスを最大化し、摩擦を排除したのである 。しかし、その経済原理は残酷な曲線を導き出した。ごく一部のアーティストが不釣り合いなほどの注目を独占する一方で、「中間層」は数百万回の再生を生計に結びつけることができず苦闘することとなった 。関心（アテンション）は、収入よりも早くスケールしてしまうのだ 。

日本の産業は異なる動きを見せた。物理メディアと国内チャートを維持することで、聴衆がグローバルなフィードへと急速に霧散するのを防いだのである 。彼らは単にフォーマットを守ったのではない。エコシステムの「密度」を守ったのだ 。その結果、構造的な緩衝材が生まれた。システムが「クリックの広さ」よりも「忠誠心の深さ」を優遇することで、より多くのアーティストがキャリアを持続できるようになったのである 。プラットフォームが「スケール」のために最適化される一方で、エコシステムは「中間層」を守り抜いたのだ 。

それはテクノロジーへの拒絶ではなかった 。それは「ペース配分（Pacing）」であった 。

市民のリズムと産業的習慣

その差異は、市場だけでなく街の風景にも現れている 。San Francisco では、イノベーションはポケットの中で鳴り響く。配車サービスは数分で到着し、食料品がドアの前に現れる 。あらゆる産業がコードによって再構築されている。しかし、物理的なレイヤー -ひび割れた歩道、逼迫した住宅事情、限界に近い公共システム——は、デジタルの輝きから切り離されているように感じられる 。ソフトウェアが、街そのものを追い越してしまっているのだ 。

Tokyo での感覚はそれとは異なる 。プラットフォームに立てば、列車は地面に描かれたマーカーと寸分違わず整列する 。静かな交差点では自動販売機が効率的に輝き、支払いは中断されることなくスムーズなタップで行われる 。コンビニエンスストアは、正確で、協調的で、特筆すべきほど当たり前の信頼性を持って、小規模なオーケストラのように機能している 。

テクノロジーは、日常生活の「振り付け（Choreography）」の中に吸収されているのだ 。その調整力（Coordination）こそが、産業的であるといえる 。Taiwan でも、その感覚は馴染み深いものだ。工場は信頼性への共通の敬意によって形作られている。サプライチェーンは絡み合い、エンジニアは国境を越えて研鑽を積み、評判は派手な演出ではなく「精密さ」によって築かれる 。信頼はゆっくりと蓄積され、慎重に使われるべきものであるという共通認識 。失敗が成功よりも長く記憶に残る産業において、規律は文化となるのである 。

これらはスローガンではなく、「習慣」なのだ 。

あるシステムは、リーチを最大化するために外へと拡張（Expand）する 。またあるシステムは、コヒーレンスを守るために内側を強化（Reinforce）する 。コヒーレンスとは単なる調和（Harmony）ではない。それはテクノロジー、制度、そして日常生活の間の「アライメント（整合）」である 。

それは、頂点が加速するときに中間層が生き残れるかどうかを決定する。イノベーションが統合（Integrate）されるか、それとも断片化（Fragment）されるかを決定するのである 。

コンシューマー向けプラットフォームはボラティリティ（変動）を許容できる。文化的エコシステムは集中を緩和できる 。しかし、テクノロジーが表面下へと潜り、鉄道システム、製造ライン、サプライチェーン、そしてエネルギー Grid（送電網）へと組み込まれるとき、アライメントは単なる美学以上の意味を持つようになる 。

ソフトウェアは速度を許容する 。 しかし、インフラは過ち（Error）を忘れない 。

Part III: グリッドが「ゲーム」になるとき

文明の整合性と適応の規律

「動乱の時代における最大の危険は、動乱そのものではない。昨日の論理で行動することである。」— Peter Drucker

「人類の本当の問題は次のことだ。私たちは旧石器時代の感情、中世の制度、そして神のような技術を持っている。」— E. O. Wilson

過去20年間、イノベーションはソフトウェアの速度で進展してきました 。私たちは Product-Market Fit (PMF) 、つまり「競合が追いつく前にソリューションをスケールできるか」に最適化してきました 。その論理が世界を塗り替えたのです 。

しかし、Peter Drucker が警告した通り、動乱の時代における最大の危険は「昨日の論理」で行動することにあります 。Artificial Intelligence (AI) は基底層（サブストレート）を根本から変えました。AI クラスタは単なるサーバールームではなく、現代の「工業炉」なのです 。そこでの成功は「デイリーアクティブユーザー（DAU）」ではなく、メガワット、冷却密度、そして稼働時間によって測定されます 。

知性が産業化するとき、パワーは構造的なものとなります 。そして、構造的システムは「ソフトウェア・クロック」では動作しません 。

Product-Market Fit から Civilization Fit へ

コンシューマー・インターネットは「可逆性」を前提としていました。バグはパッチで修正し、戦略はピボット（転換）すればよかったのです 。資本は変動を吸収できました。なぜなら、その影響はスクリーンの中に収まっていたからです 。しかし、エネルギー・システムに「バージョン管理」は存在しません 。

バッテリーの故障は連鎖（カスケード）し、グリッドの不均衡は外部へと波及します 。熱暴走（Thermal runaway）に「リフレッシュ・ボタン」はないのです 。テクノロジーが生活の基盤に深く入り込むほど、即興の余地は狭まります 。

これはスピードに対する非難ではありません。スピードは素晴らしい企業を築き、世界を繋げました 。これは、Product-Market Fit から Civilization Fit への移行なのです 。

• Product-Market Fit が問うのは「これはスケールできるか？」です 。

• Civilization Fit が問うのは「このシステムは、それが支える世界を壊すことなく統合できるか？」です 。

一方は「リーチ」を最適化し、他方は「アライメント（整合）」を最適化します。整合性がなければ、規模（スケール）は単に脆弱性を加速させるだけです 。

The Coherence Gap（整合性のギャップ）

E.O. Wilson はかつて、人類の苦境をこう指摘しました。私たちは「旧石器時代の感情、中世の制度、そして神のごときテクノロジー」を併せ持っているのだと 。この不均衡は道徳的なものではなく、構造的なものです 。

テクノロジーの進歩は、それを吸収するために設計された制度よりも早く進みます。コンシューマー市場において、そのギャップは「創造的破壊」を生みますが 、インフラにおいては「不安定さ」を生みます 。

課題は技術的な野心ではなく、Institutional Coherence（制度的整合性） です。つまり、私たちの「神のごとき」能力と、それを維持すべき規制・産業・文化的システムとの間の整合性です 。

日本の工業的気質

日本の戦後産業の歩みは、異なるテンポを提示しています 。

Sony は執拗なまでの品質管理で家電を洗練させました。Nintendo は流行を追うのではなく、規律によってエコシステムを守りました 。鉄道システムはスローガンではなく、秒単位の精度で稼働しています 。

消費者ブランドの背後には、別の層が存在します。電車、航空機、産業機械、そしてエネルギーインフラの内部を支えるメーカーたちです 。これらのセクターにおいて、失敗は単なる「四半期の業績悪化」ではなく、何十年も語り継がれる「長い記憶」となります 。

信頼は抑制を通じて蓄積され、評判は信頼性を通じて複利のように積み上がります 。これはイノベーションへの抵抗ではなく、キャリブレーション（調整）なのです 。

二つの論理の出会い

私たちは今、二つのテンポの交差点にいます 。

AI は演算能力の急速な拡大を求め、それは前例のない電力貯蔵ニーズを突きつけます 。貯蔵には、安全性、冗長性、そしてライフサイクル全体の規律が求められます 。

片側には Logic of Scale（規模の論理） があります。モジュール式システム、構成可能性、そして展開を加速させる「高速クロック」です 。

もう片側には Industrial Memory（産業の記憶） があります。エンジニアリング・レビュー、合意形成プロセス、そして長期検証を行う「低速クロック」です 。

Cold Electric が日本で Yuasa と共に歩み始めたとき、その融合は表面的なものではなく、構造的なものでした 。Yuasa は1世紀以上にわたる信頼の蓄積を象徴しています。彼らのバッテリーは、エラーが許されない新幹線や航空宇宙プラットフォームに電力を供給しています 。彼らの評判は速度ではなく、耐久性（エンデュランス）によって築かれました 。

Cold Electric はソフトウェア時代に誕生しました 。

私たちはモジュール式の Lithium Iron Phosphate (LFP) システムとコンテナ型エネルギーユニットを設計しています 。ソフトウェアのように統合され、インフラのように機能するバッテリー・アーキテクチャです 。私たちは構成可能性や急速な負荷増大、そして AI 時代の電力需要の緊急性を理解しています 。

しかし、日本でエネルギーインフラを構築することは、Silicon Valley のテンポをそのまま持ち込むことではありません。それは「適応」を意味します 。

戦略としての適応

適応とは遅さに屈することではなく、Fit（適合） へのコミットメントです 。

Silicon Valley では加速がデフォルトですが、日本では整合（アライメント）が前提条件です 。コンセンサス（合意形成）は官僚的なハードルではなく、不純物を取り除く「濾過システム」なのです 。

長時間の会議は非効率なのではなく、1メガワットを投入する前にリスクを圧縮するための「整合メカニズム」です 。Cold Electric の役割は、モジュール化されたインテリジェンスを、こうした組織的なリズムと整合させることです 。リスクが限定的な場面では迅速に、しかしリスクが構造的な場面では極めて慎重に動かなければなりません 。

規模（スケール）は依然として必要ですが、整合（コヒーレンス）は交渉の余地のない必須条件です 。

綻びなきスケール

現代の誘惑は、コンシューマー・インターネットの熱狂を再現することにあります。あらゆる犠牲を払って構築し、展開し、拡張することです 。しかし、メガワットはダウンロードではありません。グリッドはソーシャルネットワークではないのです 。

イノベーションが文明に深く根ざすほど、整合性が重要になります。整合性のない規模は分裂を招き、規模のない整合性は停滞を招きます 。エネルギーインフラの未来には、ソフトウェアのモジュール性と重工業の規律の両方が不可欠です 。私たちは自らのテンポを、目の前の仕事がもたらす結果の重みに合わせる必要があります 。

文明の底流を流れる時計

すべての文明は、異なる時間を刻んでいます 。

加速を優先するものもあれば、整合を優先するものもあります 。重要なのは、どの時計が正しいかではなく、自分が触れている「層（レイヤー）」をどの時計が支配しているかです 。

エネルギーは究極の基底層であり、あらゆるものの下に位置しています。それは現代の AI 、交通、通信の原動力です 。Cold Electric Japan は、単なるスローダウンを提唱しているのではありません。それは、テンポと結果、規模と安定性、そしてイノベーションと組織的能力を一致させる「キャリブレーション（調整）」の提唱なのです 。

テクノロジーが基盤となるとき、忍耐はノスタルジーではありません。それは「アーキテクチャ（設計）」そのものです 。私たちのイノベーションの最終的な試金石は、どれほど速く成長したかではなく、自らが変えていく世界の中で、共存できるほど安定しているかどうかにあるのです 。

Civilization Coherence and the Discipline of Adaptation

“The greatest danger in times of turbulence is not the turbulence; it is to act with yesterday’s logic.” — Peter Drucker

“The real problem of humanity is the following: We have Paleolithic emotions, medieval institutions and godlike technology.” — E. O. Wilson

For two decades, innovation moved at the speed of software.

We optimized for Product-Market Fit: Does the solution scale before the competition catches up?

That logic reshaped the world.

But as Peter Drucker warned, the greatest danger in times of turbulence is acting with yesterday’s logic.

Artificial intelligence has fundamentally changed the substrate. AI clusters are not just server rooms; they are industrial furnaces. They do not measure success in “Daily Active Users,” but in megawatts, cooling density, and uptime.

When intelligence becomes industrial, power becomes structural.

And structural systems do not operate on a software clock.

From Product – Market Fit to Civilization Fit

The consumer internet assumed reversibility. A bug was a patch; a strategy was a pivot. Capital absorbed volatility because the consequences were contained within the screen.

Energy systems are not versioned. A battery failure cascades; a grid imbalance ripples outward. Thermal runaway does not have a “refresh” button.

The deeper technology embeds into the foundation of our lives, the narrower the margin for improvisation becomes.

This is not an indictment of speed. Speed built extraordinary companies. It connected the world.

This is the shift from Product-Market Fit to Civilization Fit.

Product–Market Fit asks: Can this scale?

Civilization Fit asks: Can this system integrate without fracturing the world it serves?

One optimizes for reach; the other optimizes for alignment.

Without coherence, scale simply accelerates fragility.

The Coherence Gap

E.O. Wilson once observed that humanity’s predicament is simple: we possess Paleolithic emotions, medieval institutions, and godlike technology.

This imbalance is not moral; it is structural.

Technology advances faster than the institutions designed to absorb it.

In consumer markets, that gap produces creative destruction.

In infrastructure, it produces instability.

The challenge is not technological ambition, but Institutional Coherence: the alignment between our “godlike” capabilities and the regulatory, industrial, and cultural systems that must sustain them.

Without coherence, scale simply accelerates fragility.

Japan’s Industrial Temperament

Japan’s postwar industrial legacy offers a different tempo.

Sony refined consumer electronics with obsessive quality control. Nintendo protected its ecosystem through discipline rather than chasing every wave. Rail systems operate with precision measured in seconds, not slogans.

Beneath the consumer brands lies another layer: manufacturers whose products sit inside trains, aircraft systems, industrial machinery, and energy infrastructure.

In these sectors, failure is not a quarterly dip. It is a long memory.

Trust accumulates through restraint. Reputation compounds through reliability.

This is not resistance to innovation. It is calibration.

The Meeting of Two Logics

We are now at the intersection of two tempos.

Artificial intelligence demands a rapid expansion of compute, which requires an unprecedented surge in power storage. Storage demands safety, redundancy, and lifecycle discipline.

On one side lies the Logic of Scale: modular systems, composability, and the “Fast Clock” of accelerated deployment.

On the other lies Industrial Memory: engineering review, consensus processes, and the “Slow Clock” of long-cycle validation.

When Cold Electric began working in Japan alongside Yuasa M&B1, the convergence was not theatrical; it was structural. Yuasa represents over a century of accumulated reliability. Their batteries power the Shinkansen rail systems and aerospace platforms where error is unacceptable. Their reputation was built on endurance, not velocity.

Cold Electric was born in the software era.

We design modular Lithium Iron Phosphate (LFP) systems and containerized energy units. Battery architectures that integrate like software while operating like infrastructure. We understand composability. We understand rapid load growth. We understand the urgency of AI-era power demand.

But building energy infrastructure in Japan does not mean importing Silicon Valley tempo intact. It means adaptation.

Adaptation as Strategy

Adaptation is not a surrender to slowness; it is a commitment to Fit.

In Silicon Valley, acceleration is the default. In Japan, alignment is the prerequisite.

Consensus is not a bureaucratic hurdle; it is a filtration system.

Long meetings are not inefficiencies; they are coherence mechanisms designed to compress risk before a single megawatt is deployed.

Our role at Cold Electric is to ensure that modular intelligence aligns with these institutional rhythms. We must move quickly where risk is contained, but move with extreme deliberation where risk is systemic.

Scale remains necessary.

Coherence becomes non-negotiable.

Scale Without Fracture

The temptation today is to replicate the consumer internet’s frenzy: to build, deploy, and expand at any cost.

But megawatts are not downloads. Grids are not social networks.

The deeper innovation embeds into civilization, the more coherence matters. Scale without coherence fractures. Coherence without scale stagnates.

The future of energy infrastructure requires both the modularity of software and the discipline of heavy industry.

It requires us to match our tempo to the consequence of the task at hand.

The Clock Beneath Civilization

Every civilization keeps time differently.

Some prioritize acceleration; others prioritize alignment.

The question is not which clock is correct, but which governs the layer you are touching.

Energy is the ultimate substrate, sitting beneath everything. It powers the AI, the transportation, and the communication of our era.

Cold Electric Japan is not an argument for slowness. It is an argument for calibration — for matching tempo to consequence, scale to stability, innovation to institutional capacity.

When technology becomes foundational, patience is not nostalgia. It is architecture.

The final test of our innovation is not how fast it grows, but whether it is stable enough to live within the world it changes.

Subscribe now

1

The press release on the Yuasa M&B and Cold Electric partnership.

Sovereignty, Ecosystems, and the Shape of the Middle

“A delayed game is eventually good, but a rushed game is forever bad.” — Shigeru Miyamoto

“If you are in a hurry, take the long way around.” 急がば回れ

When mobile gaming surged, I remember sitting in a venture office in Silicon Valley as a new fund was forming around a simple, confident thesis: the smartphone would absorb the entire gaming industry.

The logic was clean: hardware in every pocket, distribution without friction, monetization reinvented. Why would anyone ever build a console again?

From that vantage point, Nintendo’s refusal to migrate Mario or Donkey Kong to the iPhone appeared stubborn, even self-destructive. Its stock price reflected the anxiety of investors who believed the wave had already passed.

The argument was consistent with everything Silicon Valley believed about scale. Platforms win. Distribution is destiny. The company that controls the interface controls the market. When a new channel opens with billions of users, you enter it — or you are left behind.

Yet, Nintendo held the line. They did not surrender their world to another company’s distribution layer. To remain still while an entire industry pivots is a rare kind of discipline. I still don’t fully understand how they held that line.

From the outside, the pull must have been immense. The iPhone had become the most powerful distribution platform in the world. Analysts circled. Investors grew restless. Every quarterly metric whispered the same message: migrate.

Markets punish patience, and shareholders rarely celebrate restraint. But restraint is often the only way to preserve coherence.

When the Switch finally arrived, it did not feel like a reaction; it felt like continuity. It carried forward the same instinct that had defined Nintendo for decades: play as experience, not feature set; delight as strategy, not specification sheet. Nintendo chose preservation over expansion — not as nostalgia, but as a strategy to protect their internal rhythm. In a world of fleeting downloads, they chose to build a contained world.

It takes a rare kind of discipline to watch a wave pass and not dive in.

The Fork in the Road: Kyoto vs. Taipei

Years earlier, I saw the “Fast Clock” version of this tension. At HTC, we built a prototype that bridged the gap between a console and a phone: Dual screens. Physical controls. Connectivity layered onto play. In short, a Nintendo DS phone powered by HTC. It was meant to add the social and communications layer to the already popular casual games of Nintendo.

I remember carrying the prototype through narrow Kyoto streets, the air cool, the architecture minimalist and precise. The boardroom was understated and deliberate, located at the end of a surprisingly nondescript lobby. There were no posters of Final Fantasy or statues of Mario to greet us; the walls were blank, offering no hint of the digital magic created within. It felt less like a gaming giant and more like a sanctuary of quiet, focused engineering.

Japanese boardrooms rarely announce enthusiasm loudly, they absorb first. The idea was met with attentive curiosity and a shared respect for hardware fluency. One of the board members leaned forward. There was curiosity. It warranted another meeting.

For a moment, it felt as though two ecosystems might meet — Taiwan’s hardware fluency and Japan’s imaginative worlds.1

But innovation is rarely decided in one room.

Back in Taipei, the prototype reached a different boardroom. The verdict was swift. The CEO disliked the usability of dual screens back in 2012. The product never shipped.

It remains one of the most disappointing ideas I helped pioneer — not because it would certainly have succeeded, but because it represented a fork in the road that was never taken.

Looking back, Nintendo’s refusal to move onto the iPhone was not stubbornness, but clarity. They understood that once you graft your world onto someone else’s substrate, you lose the power to dictate your own tempo. Sovereignty, once chosen, is rarely partial.

The Hollow Middle vs. The Dense Ecosystem

Music revealed a quieter version of the same tension.

Before streaming became synonymous with Spotify, Asia had already experimented with subscription models. I invested early in KKBox, one of the first legal streaming platforms. The model worked. Users paid monthly. Piracy declined. Catalogs expanded.

It felt inevitable that streaming would win.

But scale has gravity.

The global streaming model optimized for reach. By aggregating attention onto a single interface, they maximized access and eliminated friction. But the economics followed a brutal curve: a tiny fraction of artists captured disproportionate attention, while the “middle” struggled to convert millions of streams into a livelihood.

Attention scales faster than income.

Japan’s industry moved differently. By maintaining physical media and domestic charts, the audience did not fragment as quickly into a global feed as quickly. They didn’t just preserve a format; it protected density.

The result was a structural buffer: more artists could sustain careers because the system favored depth of loyalty over breadth of clicks. The platform optimized for scale; the ecosystem preserved the middle.

It was not resistant to technology.

It was pacing.

Civic Rhythm and Industrial Habits

The difference is visible not just in markets, but in the streets.

In San Francisco, innovation hums in your pocket. Ride-hailing arrives in minutes. Groceries appear at the door. Entire industries are reassembled in code. Yet the physical layer — cracked sidewalks, housing tightens, strained public systems — can feel detached from the digital brilliance. The software outruns the street.

In Tokyo, the sensation is different.

You step onto a platform and the train aligns precisely with the painted markers on the ground. Vending machines glow with efficiency at quiet intersections. Payment happens with a soft tap and no interruption. Convenience stores operate like small orchestras — precise, coordinated, unremarkable in their reliability.

The technology has been absorbed into the choreography of daily life.

That coordination is industrial.

In Taiwan, that feeling is familiar. Factories are shaped by a shared respect for reliability — supply chains intertwined, engineers trained across borders, reputations built on precision rather than spectacle. A shared understanding that trust is accumulated slowly and spent carefully.

In industries where failure is remembered longer than success, discipline becomes culture.

These are not slogans; they are habits.

Some systems expand outward to maximize reach.

Others reinforce inward to protect coherence.

Coherence is not harmony. It is alignment — between technology, institutions, and daily life. It determines whether the middle survives when the top accelerates. It determines whether innovation integrates or fragments.

Consumer platforms can afford volatility. Cultural ecosystems can buffer concentration.

When technology moves beneath the surface — into rail systems, manufacturing lines, supply chains, and energy grids — alignment becomes more than an aesthetic.

Software tolerates speed. Infrastructure remembers error.

Subscribe now

This discussion continues in Part III of Innovation, Fast and Slow.

1

The Nikkei report on the Yuasa M&B and Cold Electric partnership.

“The hallmark of intuitive thinking is that it occurs quickly, effortlessly, and automatically.” — Daniel Kahneman

“Software is eating the world.” — Marc Andreessen

“After the frenzy comes the synergy.” — Carlota Perez

Civilization is built on conflicting clocks.

I learned one of them in Silicon Valley.

There, speed was not a personality trait; it was a structural requirement. In networked markets, hesitation compounds disadvantage faster than error compounds risk. If a product was 70% ready, we shipped. If it broke, we patched.

Reid Hoffman’s “blitzscaling” wasn’t just a buzzword, it was an ecological adaptation. Market share, attention, and capital all follow the same compounding curve.

Marc Andreessen distilled it even more sharply: software eats the world. When distribution collapses into code and marginal cost approaches zero, the winner is often the company that scales first and fastest.

Even decision-making philosophy was redesigned around velocity. Jeff Bezos distinguished between reversible and irreversible decisions, urging teams to move fast when mistakes could be undone. Acting with 70% of the information was often preferable to waiting for 100% and losing the window.

In this environment, failure is survivable. Code is ephemeral. Servers are redeployed. The system rewards acceleration because the consequences of a bug are rarely existential.

But Tempo Is Not Speed

Years later, running a professional basketball team, I began to understand something subtler. Speed and tempo are not the same thing.

Young teams often mistake motion for control. They sprint until they collapse, resulting in turnovers and fatigue. Acceleration without rhythm.

Then there are players like Luka Dončić. He is rarely the fastest player on the floor, yet he dictates the game. He pauses when others rush. He accelerates when defenders relax. He doesn’t race the clock, he manipulates it. He dictates tempo.

Tempo is about deciding when to move quickly. Innovation works the same way.

The Silicon Valley clock rewards velocity. The arena it competed in rewarded first-mover advantage and network effects. In that arena, acceleration compounds.

But not all arenas are governed by the same clock. Other arenas reward integrity.

The Consumer Era: From Object to Platform

This was visible long before artificial intelligence and megawatt-scale data centers dominated headlines. We saw this shift in the transition from the Walkman to the iPod.

Sony built the Walkman as an object of craft – a masterpiece of hardware condensed into something you could slip into a jacket pocket. It carried with it not just music, but identity. It felt personal.

But with the iPod, Apple understood that the value had migrated from the object to the platform. Under Steve Jobs, Apple built a system: device, software, distribution, storefront. iTunes synchronized your library. The App Store later synchronized entire industries. Together, they were not just software; they were ecosystems.

The value migrated from object to network.

This was not a failure of engineering. It was an economic transformation. When distribution becomes instantaneous and marginal cost approaches zero, the center of gravity shifts. Scale becomes strategy. The companies that understood this built platforms. The companies that clung to hardware alone struggled.

Speed became decisive.

Batteries and the Logic of Industrial Expansion

A similar story unfolded in energy storage, as the logic of the platform eventually bled into hardware.

Panasonic carried forward Japan’s tradition of industrial discipline – quality control, process rigor, incremental refinement. Its batteries powered generations of devices and electric vehicles. The engineering was careful, precise, methodical.

But the electric vehicle revolution created a new competitive terrain. Manufacturing scale, supply chain integration, and capital access began to matter as much as chemistry.

When CATL expanded aggressively, beginning to build gigafactories at extraordinary speed, they didn’t just iterate on chemistry; they blitzscaled capacity. Contracts multiplied. Market share compounded.

They realized that in a global energy transition, scale is protection. Expansion is insulation.

The logic of blitzscaling moved from software into hardware. The clock quickened.

Fast and Slow as Environmental Fit

It is tempting to frame fast and slow as opposites. They are not; they are expressions of environmental fit.

Fast innovation optimizes for expansion. It assumes reversibility – that strategy can be rolled back or a bug can be patched. It maximizes reach by tolerating volatility.

Slow innovation, by contrast, optimizes for alignment. It assumes interdependence and filters risk before deployment because it maximizes for integration.

The distinction is structural, not philosophical. Fast innovation flourishes where failure is an update; slow innovation thrives where failure is a catastrophe.

In consumer apps, a misstep is a minor inconvenience. In supply chains, it is a recall. In energy systems, it is an outage.

Tempo must match consequence.

The Hidden Assumption of Speed

Silicon Valley’s confidence in velocity rests on an implicit, often unexamined assumption: that most mistakes are survivable.

If a feature fails, it is rolled back. If a market contracts, capital is redeployed. The system absorbs these shocks because the underlying foundations – electricity, bandwidth, logistics – are stable. But the margin for error narrows as innovation moves beneath the interface.

When technology begins to power hospitals, transportation networks, factories, and AI clusters consuming entire cities’ worth of electricity, speed without alignment ceases to be an advantage. It becomes fragility.

From Frenzy to Deployment

Carlota Perez describes technological revolutions in two distinct phases.

First comes Installation, a period of frenzy, speculation, and rapid buildout where capital races ahead of institutions.

The last two decades were defined by this phase. Software scaled with astonishing velocity, and interfaces consumed industries.

But we are now entering Deployment, the phase where technology reshapes the core of society.

Artificial intelligence is no longer just a digital layer; it is an industrial consumer of power. Data centers are not merely servers; they are massive industrial facilities. The conversation has shifted from user growth to megawatts, cooling systems, supply chains, and grid stability.

When technology shapes the physical world, frenzy must yield to integration.

The Clock Beneath the Screen

The consumer internet allowed us to believe that everything could, and should, move at software speed. While that belief is correct within the digital domain, it ignores the physical constraints beneath the interface.

Power grids. Manufacturing plants. Raw material corridors. These systems operate on a different clock.

They are not resistant to innovation, but they cannot absorb volatility. Their costs are measured in decades; their failures are remembered for generations.

The past twenty years rewarded velocity and scale. The next phase demands something more difficult: calibration. Apps tolerate error. Grids do not.

When innovation moves from the screen to the megawatt, tempo stops being an aesthetic choice. It becomes a structural mandate.

Subscribe now

This discussion continues in Parts II and III of Innovation, Fast and Slow.

“The container was not just a simple box. It was the key that unlocked the modern global economy.” — Marc Levinson

“We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” — Roy Amara

“The medium is the message.” — Marshall McLuhan

The most consequential technologies are hiding in plain sight. They don’t arrive as breakthroughs; they arrive as protocols. They look like boring, bureaucratic consensus.Until a decade later, when you realize the world has quietly re-engineered itself to fit their dimensions.

The shipping container did not make headlines and did not just move goods; it subsidized distance. Software containers didn’t just run code; they decoupled logic from hardware.

Today, energy is approaching its “container moment”—the point where a fragmented, bespoke system finally acquires its standard primitive.

Marc Levinson’s The Box is often cited as a history of shipping but that undersells it. The book is actually a study of how the world transforms when a messy, manual system adopts a universal abstraction.

Before1956, global trade was “break bulk”; it existed, but was slow, expensive, corrupt, and fragile. Cargo was handled piece by piece by armies of men in expensive, high-friction handoffs. Ships were liabilities the moment they docked, spending more time at port than at sea. Distance was not merely inconvenient—it was economically prohibitive.

Trade remained local not for lack of ambition, but because coordination costs were too high.

The container succeeded because it was a perfect abstraction. It didn’t matter if the box held silk or steel; to the crane and the hull, it was simply a unit of 2,400 cubic feet. By standardizing the interface, the box collapsed labor, time, and uncertainty into a single, predictable variable. It turned logistics into a platform.

Once goods were containerized, the world reorganized. Manufacturing migrated, supply chains tightened, and the modern economy became structurally possible.

Levinson’s deeper lesson is subtle but profound: the most transformative technologies are not products, but standards. They don’t win because they are flashy. They win because they change incentives and force entire ecosystems to reorganize around a new unit of organization.

Energy today—especially renewable energy—is where shipping was before the container.

We talk about solar panels, wind turbines, and grid upgrades as if the missing ingredient is scale. But this repeats the pre-container mistake. Before containers, you could always add more dock workers or bigger cranes. It didn’t solve the system. What solved the system was changing the unit of movement.

Renewable energy today is still in its break-bulk era.

We treat every project as a unique engineering miracle. Interconnections are stalled by years. Storage deployments are artisanal one-offs, requiring site-specific HVAC, fire suppression, and civil engineering. Reliability is achieved through redundancy and fossil backup rather than architecture. Energy assets are tightly coupled to geography, regulation, and custom engineering. You don’t “deploy” energy today; you negotiate it into existence.

Even storage, the obvious bridge between intermittency and reliability, has struggled to scale in a way that feels inevitable. Not because batteries don’t work, but because storage has lacked what shipping once lacked: standardization, safety, and trust.

This is not a chemistry problem; it is an architecture problem.

Our grid is a legacy monolith designed for centralized control and predictable, one-way flows. But the modern world, driven by AI, climate volatility, and edge demand, is breaking those assumptions. Just as Docker decoupled software from the underlying server, we must decouple energy capacity from the underlying geography of the grid.

This is where the analogy to software becomes unavoidable.

Before modern cloud computing, software was tightly bound to hardware. Code depended on specific servers, operating systems, and configurations. Every deployment was fragile. Innovation was slow because everything had to be coordinated end-to-end.

Then software containers arrived.

Docker didn’t make compute faster. It made compute modular. By packaging code and its dependencies into a standardized container with a clear interface, applications could run anywhere without being rewritten. Data centers didn’t disappear—but they became interchangeable. Innovation moved to the edges because the infrastructure stopped being the bottleneck.

Shipping containers made logistics composable.

Software containers made compute composable.

Energy is next.

Composable hardware is the inevitable endgame. It moves us from “custom infrastructure” to “standardized units” with known footprints, known thermal profiles, and known safety risks. This isn’t just a “better battery.” It is a new primitive: a deployable product where power goes in and power comes out, with no custom engineering required.

Once energy becomes composable, the logic of the city reboots.

Factories no longer need to cluster only where the grid is strong. Data centers no longer wait years for upgrades. Microgrids stop being artisanal projects and start behaving like repeatable systems. AI inference moves closer to where data is generated. Resilience becomes local by default rather than heroic in crisis.

The city itself changes.

Cities today are designed around centralized constraints, growing only as fast as utilities can dig trenches and upgrade substations. Composable energy turns the city into a plug-and-play platform. Data centers, EV hubs, and industrial nodes stop being “lumpy,” multi-year bets and start behaving like modular upgrades. The city stops being a rigid cathedral and begins to function like a codebase—scalable, iterative, and resilient by design.

This does not eliminate utilities any more than Docker eliminated data centers. It changes their role. The grid remains the backbone. Containerized energy becomes the abstraction layer above it—the layer where innovation happens.

This transition won’t happen because of policy; it will happen because it is the only way to allocate capital in a volatile world. Standardized, safe, and rapidly deployable assets are more bankable than bespoke ones. They turn energy from a systematic constraint into a localized platform for innovation.

And once energy is a platform, applications explode.

Not all at once.

Not predictably.

But inevitably.

That is the lesson of The Box.

That is the lesson of cloud computing.

This is the thesis behind Cold Electric: creating the physical standard that makes energy as liquid as data. It isn’t just a box; it’s the hardware API for a world that can’t afford to wait for the grid.

Subscribe now

I was sitting on the boards of a dozen companies, including Scopely, which would soon exit for billions. From the outside, the move looked irrational: leaving one of the world’s most successful venture platforms to build what many dismissed as a niche “blockchain phone.”

But for me, it wasn’t a departure. It was continuity.

I wasn’t leaving venture capital. I was following a thesis.

That thesis is simple, old, and increasingly urgent: power must be decentralized.

Not as an ideology or a slogan, but as an architectural principle—across money, identity, data, compute, and now energy. It is the same vision articulated decades ago in The Sovereign Individual: as technology advances, concentrated power becomes brittle and dangerous, while individuals equipped with the right tools become more resilient, more free, and ultimately more responsible.

The First Attempt: Digital Sovereignty (and What It Taught Me)

Bitcoin was the first credible proof that this thesis could work.

Owning your keys meant owning your money—no banks, no intermediaries, no permissions. The natural extension seemed obvious: if cryptographic keys could secure value, they could also secure identity and anchor data ownership, eventually shaping a more ethical future for compute and AI.

I didn’t discover Bitcoin in isolation. I first encountered it seriously through Horizons’ investment in Blockstream, which placed me alongside people like Adam Back. Being an investor gave me access to the protocol-level thinking that treated Bitcoin not as an application, but as infrastructure—designed to survive adversarial conditions and institutional failure.

What struck me later is that Blockstream was never really about money alone—it was about building infrastructure that could operate safely, predictably, and independently at the base layer, the same architectural problem that energy infrastructure now faces as power demand, adversarial conditions, and systemic fragility all increase.

Later, through conversations with Vitalik Buterin and deeper exposure to Ethereum, that perspective widened. Ethereum made explicit something Bitcoin implied: that cryptographic ownership could extend beyond money into identity, coordination, and computation. But it also revealed that the hardest problems begin not with code, but with human behavior.

The Lesson of the Exodus

The HTC Exodus was the manifestation of this thinking.

• Technically, it worked.

• Philosophically, it was coherent.

• But in practice, it hit a wall.

I learned—slowly and uncomfortably— that decentralization fails when it asks too much, too early. Self-custody demanded responsibility where platforms offered ease. Ownership required friction where centralized systems offered convenience, speed, and social gravity. The cost of giving up control was abstract and delayed; the benefit of reclaiming it was principled but distant.

People didn’t reject sovereignty. They deferred it.

There’s an irony here that’s worth naming: the HTC Exodus wasn’t a commercial failure. We sold hundreds of thousands of devices, and in a very real sense, it became one of the most profitable phones ever made.

Exodus was the first—and for a long time, the only—mobile phone that could natively accept Bitcoin and Ethereum as payment. At the time, that felt like a symbolic gesture toward the future of money. In retrospect, it became something else entirely. As the value of BTC and ETH appreciated, those payments compounded. What began as an experiment in sovereignty turned into a product with an extraordinary profit profile—arguably the highest-margin phone HTC ever shipped, and possibly one of the highest-margin consumer electronics products of its era.

And yet, that success only reinforced the deeper lesson: People were willing to pay with Bitcoin, but they were not willing to live with sovereignty.

Exodus succeeded commercially. The vision failed socially because the “user tax” of decentralization was too high.That distinction matters because it forced a harder question: when does decentralization actually work?

Why Energy Changes the Math

Energy doesn’t live in theory; it shows up as a monthly bill. And that bill is rising—not because of household waste, but because we are quietly rebuilding the physical substrate of civilization.

AI data centers. Electrification. Grid hardening. Redundancy. These are real, capital-intensive costs that are increasingly socialized. Everyone pays, whether they benefit directly or not.

This is where the friction of decentralization disappears.

For the first time since the Industrial Revolution, individuals can economically justify owning their own means of energy production. Solar and batteries have crossed a threshold where the math finally works—not for ideological reasons, but for balance-sheet ones.

Solar produces energy at near-zero marginal cost. Batteries allow you to move that energy across time, avoiding peak pricing and grid volatility. With a finite upfront investment, households and businesses can materially reduce—or even eliminate—their electricity bills.

This is decentralization with immediate incentives.

The Return of Energy Ownership

For most of human history, energy was local. You gathered wood; you controlled the fire. It was inefficient but direct. The Industrial Revolution inverted this. Coal, oil, and later gas required scale—mines, pipelines, and refineries. Energy became abundant, but ownership vanished. Production moved far away. Control consolidated into utilities and states. Energy faded into the background, invisible and taken for granted.

Solar and batteries reverse that logic.

For the first time in over a century, individuals can once again produce, store, and manage their own energy—this time with modern reliability and compelling economics. This isn’t nostalgia. It’s a structural return enabled by technology and forced by cost.

The philosophy hasn’t changed. The incentives finally have.

Why Cold Electric Exists

Cold Electric exists because decentralization only works when it respects reality—technical, economic, and human.

Energy is unforgiving. Unlike software, it cannot fail quietly. Unlike platforms, it cannot hide fragility behind user experience. That’s why the path toward individual energy ownership doesn’t start at the household; it begins where the demands are highest and the margins for error are smallest: utilities, data centers, and industrial sites.

These environments force discipline and safety. They force us to build systems that actually work.

We are starting where energy stress is already visible—where AI workloads are driving unprecedented demand, where grids are strained, and where reliability is existential rather than aspirational.

But the direction is clear.

As we scale, costs fall. What begins as infrastructure for enterprises gradually becomes infrastructure that individuals can own—not as an ideological promise, but as a consequence of scale.

This is how decentralization becomes real: not by starting with the individual, but by ending there.

The Bottom Line

The through-line of my career is now clear:

• Keys taught us that ownership matters.

• Identity and data taught us that convenience beats ideology.

• Energy teaches us that incentives decide everything.

Decentralization doesn’t fail because it’s wrong. It fails when it asks people to sacrifice without reward. Energy reverses the bargain by offering ownership with an immediate, measurable return.

Cold Electric is where the long arc of decentralization lands when idealism meets thermodynamics.

This time, people don’t need convincing. The spreadsheet does the arguing.

Subscribe now

Whenever I host dignitaries, government officials, and tech entrepreneurs in Taiwan, they ask the same question.

“Do you feel safe here?”

“Do you feel the threat from China?”

The honest answer is complicated. Taiwan has lived under existential threat for decades. Fighter jets cross the median line. War games run on schedule. Headlines flare up, then fade. And yet daily life continues with remarkable normalcy. Kids go to school. Factories run 24/7. Engineers ship the most advanced chips in the world.

That contradiction is the point.

As my friend historian Niall Ferguson has argued, the more likely future isn’t World War III, but Cold War 2.0—a prolonged period of pressure, blockade, coercion, and economic constraint rather than open conflict. Less invasion, more strangulation. Less spectacle, more endurance.

Adam Tooze's LRB Lecture

Adam Tooze gives this scenario its deeper structure. In his framing, the new Cold War isn’t ideological. It’s infrastructural. It’s not about democracy versus authoritarianism. It’s about energy systems, supply chains, and the material foundations of power.

Taiwan sits at the center of this reality. It’s a democratic island embedded in the world’s most critical technology ecosystems. It survives not by military dominance, but by being indispensable. TSMC didn’t just build chips—it made Taiwan impossible to bypass.

In Tooze’s terms, Taiwan matters because it sits inside the arteries of the modern electro-industrial world. And in Cold War 2.0, control over those arteries matters more than rhetoric.

Civil War 2: America’s is an Electrostate in Denial

Tooze introduces a distinction that explains almost everything happening right now: electrostates versus petrostates1.

Electrostates are not “green.” They are societies that consume more of their energy as electricity and integrate that electricity into industry, logistics, planning, and governance. Their power compounds. Petrostates derive power from hydrocarbons, extraction, and chokepoints. Their power defends.

Here’s the uncomfortable truth Tooze forces us to confront: America is already an electrostate.

One of Adam Tooze’s most uncomfortable observations is about fracking. It is often framed as a triumph of American innovation—and in a narrow sense, it is. Fracking delivered cheap energy, geopolitical leverage, and short-term resilience. But Tooze’s point is sharper: fracking didn’t transform America’s energy system; it extended the life of its petrostate culture. It reinforced existing political coalitions, infrastructure assumptions, and mental models about power. Instead of forcing a full commitment to electrification, fracking delayed it—buying time, but also locking in dependence. In that sense, fracking wasn’t a bridge to the electrostate future; it was a detour that made the transition messier and more politically fraught. It solved yesterday’s problem so well that it postponed the urgency of building tomorrow’s system.

Its real power runs on electrons—data centers, AI, finance, defense systems, semiconductors. But culturally and politically, America still thinks like a petrostate. Oil. Gas. LNG. Pipelines. The mental model of the last century still dominates the narrative.

What you’re watching in the U.S. is not a smooth transition—it’s an internal conflict. A quiet civil war over energy identity.

Under Biden, the Inflation Reduction Act and infrastructure investments pushed aggressively toward electrification. Under Trump, the pendulum swings back toward fossil fuels and energy dominance through hydrocarbons. Same country. Two incompatible futures.

So America lives inside a contradiction: structurally an electrostate, culturally a petrostate.

This is where Taiwan’s role becomes interesting again.

Just as TSMC became indispensable to America’s technology stack, the next strategic question is whether Taiwan can play a similar role in America’s energy transition. Taiwan didn’t survive by ideology. It survived by execution. By building the infrastructure others neglected.

The question is no longer abstract: Taiwan stabilized America’s chip supply chain. Can it help stabilize its energy infrastructure?

AI Changes the Rules of the Energy Game

This is where Adam Tooze’s argument becomes unavoidable.

AI isn’t just another software cycle. It fundamentally reshapes energy demand. AI data centers consume electricity at a scale that breaks the assumptions of the old grid. Single facilities draw as much power as small cities. These loads don’t flex. They don’t wait. They don’t tolerate instability.

Tooze’s point is that the new Cold War will be decided not by innovation alone, but by who can deploy infrastructure at scale. AI turns electricity into the binding constraint of economic power.

The centralized grid was built for a different century—predictable demand, one-way flow, and fossil baseload generation. AI breaks those assumptions. It requires localized, resilient, always-on power closer to the point of use. This forces a structural shift toward energy storage, behind-the-meter systems, and microgrids as core components of modern industrial infrastructure.

This isn’t climate tech. It’s electrostate infrastructure.

This is the context in which Cold Electric operates. We treat energy storage as infrastructure, not as a commodity or a financial abstraction. Our systems are designed for real environments, under real constraints, over long operating lives—safe, modular, and deployable to support data centers, factories, campuses, and critical facilities where grid assumptions increasingly fail.

Tooze warns that innovation alone is not inherently progressive. Fracking showed how technological brilliance could entrench the fossil system rather than dismantle it. We take that warning seriously. The question is not whether we innovate, but which energy system our innovation locks in.

Cold Electric is deliberately building technologies that accelerate electrification rather than prolong dependence on hydrocarbons. In a world defined by Cold War 2.0, AI, and energy scarcity, that choice is strategic.

Taiwan’s geopolitical position, America’s energy identity crisis, and AI’s explosive demand for power converge here. The companies that matter won’t be the loudest ones. They’ll be the ones quietly delivering electrons where they matter most.

If Adam Tooze is right—and the new Cold War is being fought over energy systems—then the most important actors may not be the loudest or the most ideological, but the most reliable. Taiwan has already played this role for the world. It became indispensable by supplying semiconductors to everyone—electrostates and petrostates alike—without trying to dictate how they were used. As AI reshapes the global economy, the same dynamic is emerging around energy. Cold Electric, built in Taiwan, is positioned not as the center of the transition, but as a critical partner to it: providing the physical energy infrastructure that AI data centers, factories, and digital systems require, regardless of political identity. If Taiwan is the Switzerland of advanced technology manufacturing, then Cold Electric aims to be the Switzerland of AI energy infrastructure—neutral, trusted, and essential to a world moving, unevenly and imperfectly, toward electrification.

The future won’t belong to those who talk best about values.

It will belong to those who build the infrastructure that makes power unavoidable.

Subscribe now

Looking back on my two decades of building mobile and venture investing, I didn’t realize the little battery we sourced from ATL (now rebranded as CATL) would be as critical and influential to the AI infrastructure and energy grid conversation today as the social media that mobile distributed to everyone. The battery, along with solar, is the key to the flourishing of AI.

In this blog, I will explore many ideas: how decentralization will save the world—not only money, but tech needs to be decentralized; how batteries and solar will follow the same trajectory as chips because of Moore’s Law; and how this is a geopolitical conversation on why and how Taiwan needs to be the Switzerland of technology for the world, providing chips, batteries, and other critical tech for global sovereignty.

I’ll tell the story of how I went from Bitcoin to batteries and how I went from manufacturing mobile phones (atoms) to investing in software (bits) to now working in AI infrastructure and energy services (watts). I’ll explore other untold stories as big as the Walkman-to-iPod story, where Elon saw this same battery and decided to put it in cars while companies like Panasonic gave up their position to CATL to become the dominant battery manufacturer in the world. I’ll write about the electro state and the importance of electrification and what role Taiwan has to play in it. I’ll deliberate from my perspective how the grid and the way it’s priced is about managing risk—social risk.

For the past two decades, I climbed two giant hills, from manufacturing atoms to investing in software and bits. I was orange-pilled in 2015 when I was at Horizons Ventures as investors in Blockstream. I think Satoshi is right that state and money need to be separated, like the state and church were divided during the Reformation. I think state and tech need to be separated, unlike the trajectory of where the powers are going now, and I believe this battery we’re building will help empower the sovereign individual. This hill of AI infrastructure and energy is the biggest hill I’ll climb here at Cold Electric, and I believe what we’re building will bring good energy for anyone building in this AI infrastructure space.

This Substack is a place for me to put together everything I’ve learned about atoms, bits, and watts—what it means to move through that transition, how infrastructure shapes power in non-obvious ways, and why some of the most important technological shifts happen quietly, before they’re widely understood.

More to come.